Nvidia: A Leader in AI Hardware and AI Software

Posted on July 15, 2022June 30, 2026 by io-fund

If you were to guess, when do you think we wrote the following paragraph?

“When a thesis is not reflected in the revenue segments yet, there are typically lower entry points and ongoing volatility. You’ll see in the technical analysis that although I could not be more bullish on this stock long-term, there is weakness in the semiconductor sector and we hope this translates to a lower entry point for our readers.

The market is also in a fierce debate between AMD, Intel, and Nvidia and is also distracted by other chips, such as Micron and NXP. In my analysis, I look for growth. How big is the market relative to how big the company is now?

You can ignore Nvidia’s gaming revenue and other segments for the main trajectory that we are focused on. Gaming is great for stability and earnings reports, but the growth will not be from gaming (a market where Nvidia is already a mature, market leader). I’m also not focused on PC sales or the CPU-powered cloud, as the first is not a growth market and the second is not the piece in the cloud stack that will accelerate future technologies.”

That was written in 2019 yet the far majority of those concerns could be stated verbatim right now. Do we care about PC sales or gaming consoles? No, although our stance is that we have to expect these concerns will affect our semiconductor positions at times. The good news for Nvidia and AMD investors is that as time goes on, the less consumer-related hardware will have an impact. The 2022 Nvidia Investors Presentation provided numbers which show in detail how consumer exposure will become less of a concern in the future for these AI heavyweights.

When do you think we wrote this analysis?

“Over the past few weeks, I have read many lagging explanations on the chip shortage – too many fabless semiconductor companies, too few foundries, automobile manufacturers paused ordering in March and didn’t prepare for the sharp rebound, tensions with China, and even a fire at the Asahi Kasei plant that specifically manufactures sensing devices for the automobile industry.

While all of these are true, the overarching issue is that the role of semiconductors has changed from a commodity to the primary accelerant of future technologies. This is because connectivity, automation, and ultimately AI, will disrupt every corner of every industry.

We saw this happen with data and cloud but now we must accelerate this to the next level for AI/ML and the common denominator is semiconductors. Automotive is only the beginning. We can add renewables to the list and even e-commerce as AR/VR and AI/ML attempt to prop up the leaders who are competitive enough to add these features first.

As a tech stock analyst, I don’t have the luxury of lagging analysis of any kind. My subscribers require (and deserve) forward-looking, and with my intense focus on semiconductor chips, I don’t think my readers are surprised that semis are under pressure due to an increasingly important role.

I have repeated (perhaps too many times) that there is no way forward without the semis. We are seeing this manifest in automotive right now, but as investors, we should get used to hearing about semiconductor shortages.

You and I can debate Palantir, Snowflake or C3.AI, for example, and the valuations or the right angle for AI/ML-driven software, but the common denominator to these companies is the need for semiconductors to drive forward AI and 5G.

Now, we add the enormous push for auto manufacturers to compete with Tesla, Apple, Lucid Motors and what we have is a bottle neck where the automotive industry filters into semiconductors.

My guess is the demand won’t be letting up for many years as we are no longer in the cyclical pattern that semis are notorious for. Instead, demand will outpace supply for years to come.

Is this a bad thing or a good thing for our stocks? As investors, we can either listen to the news or listen to management. In this case, they are not aligned. Machines trade off news and natural language processing (NLP) but as human investors, we have the advantage of looking deeper into the issues.

I have written volumes of analysis leading up to the triple-digit growth we are seeing now in the data center from AI accelerator chips. Most of this was written when data center growth was negative. For instance, my Nvidia thesis was set end of 2018 — and in 2019 Nvidia reported negative data center revenue year-over-year for four quarters in a row.reported negative data center revenue year-over-year for four quarters in a row.

I mention this because following a trend’s trajectory is more important than immediate gratification from the market. The trend will always win out over time.

I have maintained that chips will eventually lead the AI market and are the best angle for investing in edge computing. I have also defended our stocks against custom silicon. Now we have the first of what I predict will be many semiconductor shortages and bullish to me.

The shortage is that there are hundreds (thousands really) of companies that rely on semiconductors. This will come to a head with AI and 5G as those who go-to-market soon with these features will have an enormous competitive advantage.”

That was written at the height of the bull market in February of 2021. My goal is to illustrate there has always been headlines to worry about for the semiconductors. We’ve firmly held these stocks and bought during dips. In the past, from 2018-2019, I focused on the GPU-powered cloud and the CUDA moat here and here. Our 2020 coverage centered on the A100 GPU which we discussed at time of launch for premium here and continued coverage on the A100 about a year later on the free side.

Here is background on the A100:

“Nvidia released the Ampere architecture and A100 GPU as an upgrade from the Volta architecture. The A100 GPUs are able to unify training and inference on a single chip, whereas in the past Nvidia’s GPUs were mainly used for training. This allows Nvidia a competitive advantage by offering both training and inferencing. The result is a 20x performance boost from a multi-instance GPU that allows many GPUs to look like one GPU. The A100 offers the largest leap in performance to date over the past 8 generations.”

Nvidia's AI Dominance Will be Propelled Forward by Software:

I wanted to go back through a bit of Nvidia’s history – what was the thesis and how did the thesis evolve? – before I go into how Nvidia will continue to dominate. In my opinion, I believe this is the most important analysis I have ever written on Nvidia because the company is changing rapidly into a software company.

The shift that Nvidia is going through has gone unnoticed and that’s to our benefit. Because we have been hell bent on finding what companies will dominate AI hardware, I’ve been asked frequently who do I think will dominate AI software (Palantir? Snowflake? Google?)

I’m prepared to give you that answer today: I believe Nvidia will be one of the biggest or perhaps the biggest AI software stack company in the world.the biggest AI software stack company in the world. The analysis below kickstarts our in-depth coverage on this new thesis — and I fully believe I will be quoting this analysis in five years from now when we check back on how the AI software thesis played out.

Before I go into semiconductor jargon where I risk losing your attention, I want to make sure our Members are fully aware that the segment where Nvidia will dominate with AI software is automotive. I am not talking about a few OEMs that trickle into a little bump in revenue. I am saying that Automotive is scheduled to become Nvidia’s number one segment – even over data centers – and to the tune of it being 3X larger than its gaming segment.

Don’t take my word for it because the CFO said exactly that (more on this below) and there is ample evidence that this is happening, which I also detail for you. Wall Street won’t be giving this the credit it deserves until 2023 at the earliest but you will hear non-stop “Nvidia Automotive” coverage by 2024-2026 as this segment ramps. I go over why those are the target dates below.

But first, let’s talk about the H100 and how this new GPU architecture will also help Nvidia lead on AI software at the enterprise level. There is plenty going on outside of Automotive that we need to cover so I kept automotive for last.

GTC Highlights: The Hopper H100 GPU

In March at GTC 2022, Nvidia announced the Hopper H100 GPU with 80 billion transistors and will be released in Q3 of this year. For reference, the A100 has 54 billion transistors. This is Nvidia’s solid attempt to keep their stake in the ground in leading high-performance computing over AMD’s Instinct MI250/250X and the newly announced MI210.

It’s easy to focus on hardware with Nvidia (and AMD) yet these companies are becoming more software-driven each year. By owning the majority of the market for AI accelerators, these two companies are afforded an opportunity to also own the software layer as a means to lower the barrier to entry for training models, deploying inference across various frameworks, and other workloads related to deep learning, conversational AI, video conferencing algorithms, and more. By supplying the hardware, these companies have natural inroads to machine learning operations (MLOps).

The H100 is the New Artificial Intelligence Infrastructure

DGX, DGX Pods and DGX SuperPods:

The H100 will power all AI and high-performance computing systems including the PCI express accelerator for mainstream servers and many H100 GPUs can be combined to power advanced AI through the following systems: DGX, DGX Pod and DGX SuperPod.

The difference between the A100 and H100 is the performance will be two to three times faster. The H100 GPUs and the DGX H100 server pods and super pods offer Nvidia the next leg-up as the company has solved an important bandwidth issue.

Hopper tackles some of the bigger issues around previous generations like speeding up algorithms by offering dynamic programming on GPUs to break down problems to simpler subproblems, boosting bandwidth by 3X with SHARP in-networking computing and Infiniband Switches, and the H100 can leverage NVLink to connect eight H100s into one giant GPU for 640 billion transistors, 32 petaflops, 640GB of HBM3, and 24 terabytes per second of memory bandwidth.

The chip is custom built by Taiwan Semiconductors with a 4nm design making it the world’s fastest 4nm GPU. The H100 has about 50% more memory and interface bandwidth than the A100. That’s 1.5X more bandwidth with the NVLink connection and PCIe 5.0 doubling the bandwidth of PCIe 4.0. The H100 will ship with support for 80GB of HBM3 memory at 3 TB/s speed.

The NVLink is now able to link together server nodes to build a data center-sized GPU. NVLink was originally designed to bypass the PCIe slot and has become an important tool for chip-to-chip connectivity, especially for high-speed operations. There is a dedicated chip called the NVSwitch which has increased the NVLink’s bandwidth. The ultimate goal is to run 32 servers with their own operating systems to run a single job. NVLink will complement the InfiniBand networking for high-performance computing and NVLink will be default for all of Nvidia’s chips, including GPUs, CPUs, DPUs and SoCs.

Where the H100 really stands apart is the leap in performance with about 3X more performance than the A100 and the H100 is up to 6X faster. The leap in performance is measured by H100’s ability to deliver up to 4,000 TFLOPS of FP8 compute, 2,000 TFLOPS of FP16 compute and 1,000 TFLOPS of TF32 compute and 60 TLOPS of general purpose FP64 compute. The A100 lacked support for FP8 compute at default whereas the H100 will leverage a transformer engine to switch between FP8 and FP16, depending on the workload.

According to Nvidia, the H100 delivers 9X more throughput in AI training, and 16X to 30X more inference performance. The company also states in HPC application-specific workloads, the H100 is 7X faster. The goal of the H100 was not only to add more transistors and make the H100 faster, but to also offer function-specific optimizations. This is achieved through the transformer engine.

The architecture aims to answer one of the bigger challenges facing superfast compute, which is that moving data into traditional servers overloads the CPU and system memory and becomes bottlenecked by PCI-Express.

By improving the bandwidth issue, Nvidia’s goal is to create more demand for their DGX Pod and SuperPod Systems, which in turn, will create more demand for their software.

The H100 DGX Pod is a 32-node, 256-GPU system. The H100 DGX Pod connects 32 DGX systems using the NVLink Switch System to scale into a super-GPU capable of 768 terabytes per second. To compare, the entire internet requires 100 terabytes per second. This results in 1 exaflop of AI computing.

From there, multiple H100 DGX Pods can connect through the Infiniband Switch to scale DGX Superpods with thousands of H100 GPUs. DGX SuperPods are turnkey systems that power enterprise AI. DGX SuperPods were also available with the A100 yet the H100 will have 6X better performance with 1 exaflop of FP8 AI performance to run trillions of parameters (more on this below).

Spectrum-4 Ethernet Platform

Perhaps one of Nvidia’s most important advancements for the H100 is the ability to attach the network directly to the GPU to avoid bottlenecks at the CPU. This is accomplished by sending data with direct memory access at 50 gigabytes per second. Hopper HGX and DGX are networking and interconnects that facilitate moving data with an advanced networking processor called the CX7. The result is the H100 CNX that avoids bandwidth bottlenecks and frees the CPU and system memory to process other parts of the application.

The Spectrum Ethernet platform, which consists of a Spectrum-4 Switch, CX7 SmartNIC and Bluefield-3 DPU will be used for several of Nvidia’s AI platforms, such as Riva, Merlin and Omniverse. These workloads include natural language processing, recommenders, and digital twins and will be supported by a networking system that helps exchange large databases between nodes. Whereas traditional workloads required many connections exchanging small amounts of data, the workloads of the future will require data to be shared quickly between GPUs and storage. This is accomplished by bypassing the CPU and sending data directly to the GPU while using the network hardware to move the data.

This is ideal for enterprise use cases where people are more likely to use Ethernet while AI and HPC workloads continue to use the Quantum-2 based off Mellanox’s InfiniBand. Quantum-2 allows for in-network computing to do data reductions in the network. It’s also more likely that Ethernet is used for 5G and sensors.

Eos: The First Hopper AI Factory

Nvidia is building AI factories to compete with AI supercomputers, which are blueprints for AI infrastructure that can be adopted by cloud partners and enterprises.

Eos is built with 18 H100 SuperPods, with 576 DGX H100 systems and 360 NVLink Switches. Nvidia states EOS is 1.4X faster than the fastest supercomputer and offers 4X the AI processing of the world’s fastest supercomputer. This will deliver 18 EFLOPS of FP8 AI compute or 9 EFLOPS of FP16 compute.

Previously, FP16 was the standard for AI whereas FP8 is gaining more support to become the industry standard. Depending on what AI compute you use, benchmarks will not be apples-to-apples if FP8 is compared to FP64 performance. Right now, AMD’s Frontier supercomputer is #1 with 1.1 exaflops of FP64 performance compared to the upcoming Venado supercomputer’s 10 exaflops of FP8 performance.

The difference is that the smaller bit size allows for an economical way to achieve more speed when giving up a small amount of accuracy doesn’t make a critical difference. This also helps in the face of a slowing Moore’s Law. FP8 is most commonly used for inference yet may be used for training in the future due to boosting throughput. Following the release of the Hopper H100, Intel released Gaudi2 which supports FP8. Chip makers Graphcore, AMD and Qualcomm have recently pushed for an industry-standard for the low precision floating point format FP8 rather than integer formats.

Here is what Nvidia said in the GTC keynote:

But the trend in AI computing has been toward developing neural nets that lean on the lowest precision that will still yield an accurate result. The smaller formats compute faster and more efficiently, and they require less memory and memory bandwidth. The addition of 8-bit floating-point units in the H100 leads to a significant speedup—double the throughput compared to its 16-bit units”

DPX Instructions (ISA):

The H100 improves dynamic programming with DPX Instructions that will help specific AI Algorithms to perform up to 7X faster than previous GPUs and 40X faster than CPU-based algorithms. As algorithms require more complexity, the new set of DPX instructions will help break the complex problems down into simpler subproblems using GPUs instead of CPUs or FPGAs.

The DPX ISA are expected to be broadly available with the CUDA 12.0 release. Examples of where this will be useful include disease research and drug discovery where the process can be sped up 35X for real-time processing to match the rate of DNA sequencing. Route optimization and finding the shortest distance between destinations for use in factories and autonomous driving systems, or Floyd-Warshall acceleration, is boosted up to 40X compared to CPU-only servers. These instructions will also be used for quantum computing and SQL queries as dynamic programming can help find the optimal order for joining a set of tables.

GPU Confidential Computing:

Data is encrypted at-rest and in-transit yet is often unprotected during use. Meanwhile, the data used to train AI models is worth millions in investments and is trained from domain knowledge and company-proprietary data. The new H100 offers confidential computing whereas previously only CPUs offered the protection of both data and applications during use.

Nvidia is Becoming a Leading AI Software Company

It would be easy to read the information above and to assume Nvidia is improving its hardware. However, the company’s future resides in software which will remove some of the cyclicality of hardware revenue. I believe once Nvidia’s software revenue begins to reveal itself in earnings reports, the market will finally piece together the true potential of this AI powerhouse.

It’s both the hardware and the software stack that led me to say Nvidia will surpass Apple in 5 years. You know this story well: the relationship between a hardware company leveraging their position to capture the lion’s share of the software — because that’s exactly what Apple did.

There are four layers to Nvidia’s full-stack accelerated computing: hardware, system software, platform software and applications. Below, I discuss a few ways that Nvidia is capturing more of the software stack due to vendor lock-in effects from their dominance in hardware.

As stated, in the past, our focus was the GPU-powered data center. This was a four-year thesis from 2018 and we doubled up on the thesis in June of 2020 for the A100 release. I want to make sure and emphasize that Nvidia’s lesser-known catalyst is actually the software.

The H100 is helpful in maintaining a lead in GPUs, which is critical turf to protect with GPUs being the most popular AI accelerator, however — the AI/ML catalyst will be further fueled by the Nvidia’s lead in software. This is why the majority of who will remain the AI leader will be up to developers and not the C-suite partnerships on hardware that characterized Intel’s lead over the past few decades. The developers choose the frameworks, the SDKs, libraries and the other parts of the software stack, and because of this, they also choose the GPUs they build on rather than IT departments.

Transformers

The transformer engine is one of the key aspects of the H100. Transformers are becoming one of the most popular neural-network models by applying self-attention to detect how data elements in a series influence and depend on one another.

Sequential text, images and video data are used for self-supervised learning and pattern recognition, which results in more data being used to create better models. Prior to transformer models, labeled datasets had to be used to train neural networks. Transformer models eliminate this need by finding patterns between elements mathematically, which substantially opens up what datasets can be used and how quickly. Transformers are partial to the parallel processing that GPUs offer.

Google first introduced transformer models in 2017 and transformers are used in Google and Bing Search. Transformers also led to BERT models, which stands for Bidirectional Encoder Representations from Transformers, and is commonly used for text sequences. Transformers are also used in GPT-3 (it’s the T in GPT) which improved from 1.5 billion parameters to 175 billion parameters. GPT-3 has the ability to report on queries it has not been specifically trained on.

Nvidia and Microsoft recently worked on a Mega transformer model with 530 billion parameters and the future for AI engineers is trillion-parameter transformers and applications. The H100 is already prepping for this. According to Nvidia, the training needs for transformer models will increase 275-fold every two years compared to 8-fold for other models. The H100 GPU with its Transformer Engine supports the FP8 format to speed up training to support trillion-parameter models. This leads to transformer models that go from taking 5 days to train to becoming 6X faster to only taking 19 hours to train.

The transformer engine is software combined with the new hardware in the H100’s tensor cores. As discussed, the A100 was designed for floating-point numbers to 16 bits while the H100 is designed for 8 bits. This is helpful because AI models are moving toward neural nets that lean on the lowest precision and yet still yields an accurate result. In this case, 8 bits double the throughput of 16-bit units, compute faster and more efficiently, and they require less memory and memory bandwidth.

The main feature from the Transformer Engine is the ability to choose what precision is needed for each layer in the neural network at each step, transitioning between 8-bits, 16-bits, 32-bits, plus the H100 is able to do matrix math with two forms of 8-bit numbers with either 5-bits as the exponent or 4-bits as the exponent: E5M2 and E4M3. This is important because the E4M3 may be favored for back propagation while E5M2 may be favored for inferencing.

Pictured above: Nvidia is prepped to support model sizes growing up to 275X every two years

Triton Inference Server:

Nvidia offers AI frameworks to reduce time for developers throughout the AI workflow from data processing and ETL to deep learning model training and large-scale inferencing. These libraries include Dali, Rapids, Triton and Magnum I/O. The library supports all popular frameworks and offers pre-trained models and data pipelines.

Triton is open-source inference software that helps developers deploy models across GPUs and CPUs, it supports Tensor Flow and PyTorch, any query type and any model – such as Transformers or CNNs (used for image recognition) and RNNs (used in speech recognition). The inference engine helps developers take AI development from experimentation to production by removing the need for multiple inference servers and simplifying machine learning infrastructure on the backend.

MLOps (machine learning operations) helps developers with less ML expertise to train and deploy models yet there were limited use cases with little help in deploying custom models. Triton offers high performance inference and scalability with Dockers and Kubernetes while serving up to hundreds of models with the model control API. By supporting all popular frameworks, Triton helps developers avoid framework lock-in due to the consistent interface regardless of training framework or hardware.

Nvidia will Power the Lion’s Share of Automotive – and that means software licensing

Nvidia’s lead in automotive across dozens of OEMs requires its own deep dive. The reason I haven’t prioritized this is because Hyperion 8 is shipping in 2024 and Hyperion 9 will ship in 2026. However, as long-term investors, we should touch base now on the long-term vision for yet another large and sweeping revenue segment. In fact, automotive promises to be Nvidia’s largest segment by 2030 – so on that alone, imagine what Nvidia investors have in front of us.

Nvidia’s Orin SoC (system-on-a-chip) is designed for the neural networks that run robots and AVs at the edge. This is the central computer for the car. The Orin SoC is capable of 254 trillion operations per second by combining Nvidia GPUs with Arm CPU cores and TensorRT APIs. The goal is to help OEMs move from Level 2 autonomous systems to the elusive Level 5 and it stiffens the competition with Tesla’s FSD. Notably, at the release two years ago, Tesla pointed towards Orin’s power consumption as a potential issue for EV batteries but this has not stopped many competing EVs from adopting Nvidia’s in-vehicle hardware and DRIVE software stack.

The EV manufacturers that have already moved forward with Nvidia DRIVE Orin include: Nio, Xpeng, BYD, Lucid Group, Mercedes and Land Rover, GM Cruise — you name it, it’s probably in production with Nvidia at this moment. The company’s current automotive pipeline exceeds $11 billion over the next six years – expect this small blip of pipeline to grow exponentially.

Nvidia recently announced an upgrade to Orin called Atlan with 1,000 TOPS on one chip, or more than then Level 5 compute in AVs today. This chip will catapult forward the computing performance of AVs and is expected to be released in 2023.

Nvidia DRIVE is the operating system and software stack for vehicles that offers an execution environment and includes both security and over-the-air updates. DriveWorks is an SDK that enables self-driving applications. Drive AV offers key ingredients to an autonomous system, such as perception, mapping and planning modules. Regarding mapping, Nvidia DRIVE Map is a multi-modal drive engine that can map independently and has two map engines. Drive IX is open-source software that offers vision, voice and graphics for the user experience. (I will do a separate deep dive on Nvidia Automotive in 2023).

The entire autonomous platform is called Hyperion, which includes the compute and sensor toolkit. This includes the hardware plus a 360-degree camera, radar, lidar and ultrasonic sensor suite. As stated, Hyperion 8 ships in 2024 with Hyperion 9 shipping in 2026, which will double the processing speed and offer an increase in sensors. Nvidia offers open-source developer kits to help increase its compatibility across various projects.

Rather than train the vehicles on the road, Nvidia trains in simulation and can create virtual world obstacles for the vehicles to learn from. This is a different approach from companies like Tesla who have millions of cars on the road collecting data which they then augment for unusual events with a photorealistic simulator.

Tesla has the most data of any car manufacturer which helps the company competitively as more data equals better performing models especially in terms of object detection. More data from millions of cars on the roads also helps with prediction as Tesla collects data from incorrect predictions that can be added to the training set. By leveraging a prediction neural network, Tesla does not need to use human labeling or annotation and can instead use what’s called a temporal sequence of events — in other words, Tesla rewinds events and labels objects automatically with the use of a supercomputer.

The advantage here is that training neural networks correlates with the miles (which again, are substantial due to size of fleet on the road compared to competitors) rather than correlating with the need for human labeling. The result of automatic labeling is that Tesla is able to predict rare situations with more accuracy.

Where Nvidia delivers a strong advantage is the company has decades of history with graphics and simulation due to its gaming roots. As stated, Tesla also uses imitation learning and has a photorealistic simulator which uses vector space for labels and functions like a game engine. However, Nvidia has been quietly working on their simulation platform for many years internally despite only recently marketing Omniverse to the public. In this case, Nvidia has such a high-level of confidence in their simulation skills that they forego the real-life fleet to primarily train virtual 3D models. The company is also packaging the simulation platform for many other uses cases, such as AI factories, 5G networks, power plants and climate research. Developers can work with 3D tools through Python-based development.

Here’s a 10-minute demonstration with the simulation platform here around minute 7:00.

To keep it simple, Tesla’s primary advantage is the data they have collected as no other EV/AV has collected this level of data from real drivers. To contrast, Nvidia has arguably the best simulation platform due to decades of graphics work. These digital twins are only now being widely marketed despite being in development for over 5 years. The license costs $9,000 and Nvidia has estimated its current addressable market is 20 million engineers. Notably, Nvidia’s Hyperion will also be deployed in millions of vehicles over time, offering similar levels of data as Tesla’s fleet.

The Tesla VS Nvidia debates have not formally begun but they are certainly in our future … so brace yourself. Ultimately, the way Nvidia stands apart is the company does not directly compete on manufacturing vehicles. This is something anyone can agree on. That means many OEMs will use Nvidia’s DRIVE system whereas Tesla is less likely to commercialize their software as they’re viewed as a main competitor.

As long as Nvidia continues to innovate and maintain a lead, the popularity of its DRIVE system is likely to remain due to the company’s strategic advantages in AI and supercomputing. The company did an excellent job of tackling the edge computing use case of autonomous vehicles first.

Hardware is only part of the equation. The long-term plan is for Nvidia to license software for autonomous vehicles, which will create a recurring revenue stream. The licensing fees will go well beyond Omniverse to include the actual owner of the vehicle paying a subscription fee to Nvidia for its software. Tesla does this with their AutoPilot software which has grown from $5,000 to $12,000 per vehicle.

The breakdown according to the 2022 Investor Presentation looks like this:

$100 billion from gaming
$300 billion from chips and systems
$150 billion from AI Enterprise software
$150 billion from Omniverse software – fees are charged to both users and robots/digital twins
$300 billion from Automotive – primarily software

What Nvidia is communicating is that software revenue will surpass hardware revenue long-term.

Here is what Kress stated: "Our software content per vehicle can be in the thousands of dollars over the lifetime of the vehicle compared to the hundreds of dollars for the hardware. And second, software scales with the installed base of vehicles, not annual production.”

Note on CUDA:

The software discussion on Nvidia is not complete without a mention of CUDA. We called this Nvidia’s moat back in 2018 and we continue to believe it provides an important moat. The CUDA-related libraries include frameworks that span quantum computing, robotics, 5G networks, cybersecurity and drug discovery. The universal skills taught around CUDA and Nvidia’s SDKs help to drive more business for Nvidia’s GPUs.

Note: I’ve covered Omniverse in-depth here.

Risk: Valuation

The primary risk right now is valuation as Nvidia trades 2X higher than its peers on both the top line sales valuations and on the bottom line with earnings and cash-based valuations. There’s probably equal risk in waiting for Nvidia to drop another 50% as there is in buying Nvidia at the 2X valuation. One reason Nvidia may be valued here is because it’s slowly becoming a software company. Regardless, Knox’s technicals help immensely in determining if the market will continue to award Nvidia it’s gold medal valuation or if the market will discount Nvidia based on sentiment-driven headlines. This is a position we plan to keep on building so you can keep an eye out for those trade alerts over the next few years.

Conclusion:

Finding great companies is only half the battle, fighting negative sentiment is the other half – and semis have no shortage of this in any market – hence our beginning quotes from 2019 and also 2021.

Nvidia is the strongest company in terms of product on the market today. That doesn’t mean semis won’t be a roller coaster – we should fully expect that semis will undulate in sentiment and price while we hold our stocks over many years. We can’t change the way Wall Street works — which is a pendulum that swings between value stocks and growth stocks — but we can describe in great detail why concerns around gaming and consumer electronics slowing down is not going to change our position. We do not care to perfectly time entries or to find a perfect bottom – you’ll be hard pressed to find any legendary investor recommend that this be an investor’s goal. What we care about is finding quality companies and building those positions over time. Nvidia fits this description.

Special Report: The New Kings Of Tech

Posted on June 6, 2022June 30, 2026 by io-fund

The FANG acronym rose to popularity in 2013 and was extended in 2017 to FAANG to include Apple. If history is any indication, the world’s most valuable companies over the next ten years will not look like the previous ten years. Being early to identify which companies can take over this coveted status is how generational wealth is built.

As a tech industry analyst who has seen what one generous winner can do for an entire portfolio, I want to pause and acknowledge that an investor needs to only identify one company that can hold a top 5 position in order to see life-changing gains. To choose all five would be to defy incredible odds. This analysis is aimed at identifying what companies we believe will hold “world’s most valuable” by 2030.

Not only is tech the most valuable industry today, but what the tech industry is setting up to do over the next ten years will provide exponential gains compared to the 2020-2030 era. With that said, tech is going through a period of consolidation and this means the stakes are high in identifying the winners. To complicate matters, the market is not efficient with tech stocks as each product is quite nuanced and impossible to efficiently price without manual deep dives. Instead, the market will indiscriminately penalize all tech and indiscriminately reward all tech — and each time the liquidity tide rolls in and then out again, it becomes sink or swim. At times like this, we are very flexible as we know we only need to identify a handful of winners.

Our portfolio has twenty positions at any given time, yet we believe it will be 5-10 positions total that create 90% of our wealth by 2030. We are long-term buy and hold investors yet we acknowledge and accept this means we exit those who show weaker-than-expected results for more than one quarter — or we trim to 1% to hold a place for the stock in our coverage. Because we are flexible, we can always revisit a stock when the story resumes and the earnings match the thesis again.

The equity market is driven by sentiment and macro factors, which we expand on herewe expand on here, yet the underlying strength of tech fundamentals is hard to deny. The best way to predict what will happen next is to look closely at what happened over the past decade.

In 2007, following Steve Jobs famous iPhone keynote, a burgeoning app economy was driven forth by iOS and Android developers. Google’s search engine was already a success yet mobile catapulted it’s use by putting the mobile device into far more hands over three years’ time than personal computers did over two decades.

There were many ways to capitalize on this massive addressable market — the iPhone and iOS apps dominated the highest spend on mobile, Facebook proliferated to 2 billion people and Google expanded to acquire YouTube. In this way, mobile drove gains for three FAANGs.

The adage is that history rhymes but it does not repeat. I believe a large addressable market is certainly required to produce the new wave of FAANGs – however, rather than consumer driving the gains, I believe it will be enterprises.

Below, I discuss the enterprise-level market that will be four times larger than mobile and two stocks that will directly participate. Imagine participating in 4X the FAANGs by 2030. That’s what I believe will happen due to one key trend and I discuss exactly why this will be achieved below.

Later in the analysis, we look at cloud and the trends that will drive cloud over the next ten years. This will catch investors who are complacent off guard as cloud is already going through a period of consolidation and we are seeing new business models emerge, such as the consumption model whereas the SaaS model with annual recurring revenue has dominated the past decade. Microsoft helps prove that cloud is certainly capable of FAANG-status.

We will also look at blockchain and crypto as I have been covering this space since 2013, which predates the Ethereum network. I was trained in Silicon Valley and my role is to introduce public investors to the next wave of innovation in as safe a manner as possible. I agree that 90% of cryptos will go to $0 yet I/O Fund has been firm for years that Bitcoin would reach $1 trillion in market cap and we have two Layer 1 networks to discuss with you plus a middleware company. Whether you’re ready for it or not, Web3 will replace the internet by 2028-2030 and we are fully prepared to participate in the substantial gains the blockchain will produce.

Lastly, FAANG is not entirely dead and consumer will have its moments too. We discuss the two FAANGs we own and what major catalysts these companies have in their future. We also briefly touch on some consumer-facing stocks we own and the large addressable markets they have the potential to capture.

For those of you who are new to the I/O Fund, we are prolific in our analysis. I began writing analysis on products, startups and enterprise-technologies in 2011 and moved over to the public markets seven years later. There is a library of analysis available to you that dates back to our launch in 2019 and additional free analysis in 2018. Due to the sheer number of products I have analyzed, we are able to hold an all-tech portfolio across semiconductors, cloud, ad-tech, blockchain and more.

We also encourage you to sign up for trade alerts as Knox’s active tradesKnox’s active trades help frame the market and whether we are risk-on or risk-off. We also have an automated hedge signal and are audited annually. You can learn more about how we manage an all-tech growth portfolio here.

Yet, the investors on our site need to do their part, which I can summarize as the following:

Speak with your financial advisor about your risk level. We are not financial advisors. Instead, we simply show you the trades we are making with our own money.
Use our pie chart to view our allocations. The larger the allocation, the higher the conviction – and vice versa, the lower the allocation, the lower the conviction.
As stated, we are flexible as we expect a handful of companies to drive the majority of our gains. If we receive new information, we will manage risk accordingly by lowering allocation or exiting the position.
We firmly believe all tech investors need a long-term time frame for tech. The best tech investors in the world are venture capitalists and they seek an exit 5-7 years after they’ve funded a round. The reason to have a long holding period is that it’s nearly impossible to time an entry, therefore professionals instead time their exit. By having a long-term horizon, you can be patient until the market conditions are in your favor to take your exit.
Accept that tech is volatile. For example, high beta tech has sold off around 40%/year since 2018. I have been down this much and more so on positions that became 1,000% and 2,000% winners. This is not value investing, it’s an entirely different sport.
We have a proprietary hedge that we developed. The hedge went live in April and is designed to help us remain comfortably invested during drawdowns. You can learn more about the hedge here and watch the webinarhedge here and watch the webinar.

Without further ado, let’s talk about who will be the world’s most valuable companies in 2030

Artificial Intelligence and Machine Learning will Exceed the Mobile Economy

Smartphones had a 10-year cycle of maturation with the iPhone distribution beginning in 2008 and the app economy had a similar 10-year maturation for digital advertising. We know mobile is reaching saturation as the iPhone often has flat quarters and Facebook’s DAUs are also flat. Following a decade-long run:

The smartphone market was valued at $720 billion in 2019 and the global mobile application size was $155 billion.
The mobile advertising market was valued at $60 billion — Facebook
The total global ad spend worldwide is valued at $560 billion — Google

The market is roughly $2 trillion for mobile yet the market cap of these companies combined is $4 trillion. Meanwhile, Pricewatershousecooper is predicting the AI market to reach $15.7 trillion with some believing AI will be the next electricity. Semiconductors will not comprise the entire $15.7 trillion but according to McKinsey, they will “capture 40 to 50 percent of the total value from the technology stack.”

“Many AI applications have already gained a wide following, including virtual assistants that manage our homes and facial-recognition programs that track criminals. These diverse solutions, as well as other emerging AI applications, share one common feature: a reliance on hardware as a core enabler of innovation, especially for logic and memory functions.”These diverse solutions, as well as other emerging AI applications, share one common feature: a reliance on hardware as a core enabler of innovation, especially for logic and memory functions.”

The artificial intelligence economy will be four times larger than the mobile economy. Put differently, mobile gave us companies with up to $2 trillion market caps and AI will give us companies with $8-$10 trillion market caps. Let’s break this down.

The size of total addressable market is critical to produce the world’s most valuable companies. FAANG companies have illustrated this well and that the private markets base nearly every investing decision on TAM.

Pictured above: Google’s search engine revenue growth from 2008-2022

Google’s search engine is used by over 4 billion people
Android is used by 2.5 billion people and YouTube 2 billion people.
Facebook is at 2.9 billion users
Apple has 1.8 billion active devices (about 1 billion iPhone)
Microsoft Windows has 1.4 billion users and MS Office has 1.2 billion users = Microsoft coming from the dot-com era shows us that mobile produced much larger addressable markets across three companies compared to the previous decade.

However, not only will AI semiconductors power the accelerated computing and the training and inference that reaches every person on earth, it will ultimately power the automobiles, the streetlights, vehicles, refrigerators, factories, cities and spaceships. This will extend the addressable market beyond the 7 billion global population to reach 100 billion connections. Now, imagine this – it will be writing the software too and running the machine-to-machine automations.

View my Bloomberg appearance here where I discuss the AI market being 4 times larger than mobile.Bloomberg appearance here where I discuss the AI market being 4 times larger than mobile.

Here is a glimpse of what AI will do for GDP in each country:

Source: AccentureAccenture

AI is expected to nearly double GDP in the United States by 2035 and across Europe and Japan. The same study shows the American worker will increase productivity by 35% due to AI.

Accelerated Computing Required for Artificial Intelligence and Machine Learning will Produce Two (New) FAANGs: Nvidia and AMD

Accelerated computing is a term first used by the gaming industry when graphic accelerators were put into use to accelerate the work of a central processing unit (CPU). Nvidia created GPUs to offload tasks from CPUs for rendering 3D images. The CPUs provide the instructions while GPUs perform multiple calculations from streams of data. Parallel processing became a natural fit for the data center including training artificial intelligence and deep learning models due to processing multiple computations simultaneously.

Please reference the additional resources below where we have done previous deep dives on every company mentioned in this summary.Please reference the additional resources below where we have done previous deep dives on every company mentioned in this summary.

Nvidia’s Moat is Called Cuda:

Nvidia has a parallel computing platform and programming model called CUDA that is universally known due to the company’s first mover advantage in GPUs. The GPUs themselves do not create the moat. The compute platform creates the moat. Universities teach CUDA, which helps strengthen the universal platform for building GPU-accelerated applications as students graduate with this universal skill.

Hardware does not offer a moat. The iPhone was not the moat. Instead, it was the strength of the developer ecosystem that propelled Apple to become a $2 trillion company. The moat that Apple has enjoyed was created by the third-party developers who created iPhone applications in C and C++ with XCode, which made the device more attractive due to the mobile app ecosystem.

Android then became the second operating system in the mobile duopoly. Due to the friction of learning too many languages, the mobile ecosystem did not entertain any further competitors. This is despite there being 4 to 5 billion smartphones globally (i.e. it’s certainly feasible from a consumer supply/demand view point to entertain more operating systems and app stores), yet the limitation came from the number of languages developers are willing to learn. Microsoft Windows failed because it launched too late, and developers had already chosen the two languages they were willing to work with.

Developers create a moat because they don’t want to learn new systems – the cost of time, especially when bringing products to market is very valuable. For instance, AI startups are not going to shop a new software layer to program GPUs right now as it’ll slow down their time to market and it’s critical to launch products quickly. If there’s a competitor to Nvidia and the startup is already developing on the CUDA platform, then it better be a heck of a value proposition.

Nvidia’s Game Changer Was the A100 GPU:

In 2019, I had already stated to our premium research customers that Nvidia would become one of the world’s most valuable companies. However, the path became clearer in 2020 when the company released the A100 GPU which combines both training and inference onto one chip. The result is a 20X performance boost from a multi-instance GPU that allows many GPUs to look like one GPU. The A100 offers the largest leap in performance to date over the past 8 generations.

Note: Reference the resources below for more information on the A100 GPU including our coverage in 2020.

Fast forward, and nearly two years later after the A100 GPU launched, Nvidia had this to say in the most recent earnings report:

“[Data center revenue] doubled year-over-year. and we're seeing really strong adoption of A100. A100 is really quite special and unique in the world of accelerators. And this is one of the really, really great innovations as we extended our GPU from graphics to CUDA to Tensor Core GPUs. It's now a universal accelerator […] And so from database queries to data processing, to extraction, and transform and loading of data before you do training and inference and whatever image processing or other algorithmic processing you need to do can be fully accelerated on A100.”

Buried a bit deeper into the previous earnings call, management stated this: “The flagship NVIDIA A100 GPU continue to drive strong growth. Inference-focused revenue more than tripled year-on-year.”Inference-focused revenue more than tripled year-on-year.”

These are the kinds of critical moves we try to get in front of by covering the A100 GPU at its launch. Two years later, and we see management saying inference revenue has tripled and data center revenue doubled due to this specific product.

View my interview on TDAmertrade where I discuss Nvidia’s data center segment and why automotive will be strong in the second half of 2022.my interview on TDAmertrade where I discuss Nvidia’s data center segment and why automotive will be strong in the second half of 2022.

AMD: The Dark Horse of AI Chips

The Dark Horse refers to an underdog or an overlooked competitor that emerges seemingly from nowhere to succeed. We believe AMD is a force of nature that the market often underestimates due Intel’s overhyped public relations strategy. Meanwhile, the competitive prowess of Lisa Su has led to the second biggest comeback in the history of the tech industry after she took over AMD in 2014 on the brink of bankruptcy.

Note: We’ve done a 1-hour webinar exclusively on AMD. Reference the resources below.

AMD’s Strength Came from the Zen Architecture

The Zen architecture was introduced under Lisa Su in 2017. These processors are chipset free and fully integrated. Communication between CPUs is done through Infinity Fabric protocols. The result of the new architecture was more energy efficiency and the ability to execute more instructions per cycle.

The company released the second generation of its Zen architecture and this is when AMD started to clearly outpace Intel in terms of computing power, memory and energy use – all at a lower cost. This was due to multi-chip modules that combine a 7nm with a 14nm to use the most advanced technology when and where it’s needed most by leveraging the more mature process node. The L1 cache and L2 cache locations in the core and across the core also helped the company beat Intel on memory bandwidth.

Intel was still producing a 14nm chip with a 10nm supposedly on the way. Essentially, AMD leapfrogged the incumbent with a product that is more power efficient and allows for more cores per chip.

Note: read more about the Zen architecture in the resources listed below.

If technical jargon around chips isn’t your thing, then this is probably the most important line from our original analysis in terms of AMD’s competitive prowess: “It’s estimated that for every $1.00 in Rome chip sales, Intel loses $2.25 on average in Intel Xeon SP sales. The savings are then deployed to buy more Rome chips, which can further depress Intel’s revenue.”

We can clearly see AMD taking market share in server CPUs although losing ground in desktops and laptops (our thesis is server market share so that’s less important to us). Notably, overall CPU market share for AMD is up.

Most importantly, look at where AMD was when it launched the second generation of Zen (roughly 2%) to today (roughly 11%) market share – or nearly 6X from this major design win. Moving forward, Intel will need to deliver a 7nm chip – but by then Lisa Su will already be releasing a 5nm design.

As the analysis below points out, Intel needs to make up for lost time, meanwhile, Lisa Su is unlikely to allow that now that AMD has clawed its way back from near-zero. Our site was early to AMD overtaking Intel and this was the analysis we chose to publish during the Covid selloff in March of 2020.

Tech investors should pay attention to AMD’s ability to stave off the competitor and the new inroads AMD will have following the Xilinx acquisition. Xilinx’s FPGAs are particularly well suited for accelerated computing yet require an easier software development platform – which I suspect AMD and Lisa Su will fully deliver in the next year.

So far, Lisa Su has simply set the foundation for her company to see substantial AI tailwinds.

AI Acceleration Goes Far Beyond Data Centers

In the months to come, I will detail for I/O Fund members the additional revenue segments that will cause Nvidia and AMD to catapult their current market caps. The data center does not even scratch the surface.

Primarily, these companies will participate in the lion’s share of AI acceleration in the automotive industry, edge computing and edge devices, 3D virtual worlds and robotics simulation, industrial automation, software automation — and probably most crucially, why the leading hardware companies of today are moving into licensing software and why that will cause an eruption in revenue for these particular hardware companies. Look for this special report before next earnings season.

Before I move onto cloud, I’d like to mention that we hold two more semiconductor positions – Marvell and Lam Research. We foresee holding these companies for the long haul and linked resources below spell out why we’ve chosen these two names out of the hundreds of semiconductors on the market.

Nvidia Resources:

AMD Resources:

Marvell Resources:

Lam Resources:

Lam Research Analysis: 2021/2022 Update

2030 Cloud Companies Won’t Look like 2020 Cloud Companies

Tech is synonymous with innovation, and consequently, innovation is synonymous with the word change. This is why winning cloud investments from the past ten years will not look like the next ten years. Cloud has treated investors well, yet cloud is going through a transformation that will shake up the previous paradigm. The previous paradigm was one where most cloud stocks were successful, and was distinguished by easy cash. Now that cash is tighter, there will be fewer winners in this category. We covered the fundamental change to cloud’s bottom line here in: Compartmentalizing Cloud Stocks.

Cloud: Only the Strong will Survive

In 2010-2021, the public markets saw hundreds of cloud companies go public. Yet anyone with a decade or more experience in tech will tell you that consolidation eventually will come knocking.

Consolidation is a natural part of the tech industry where competitors become acquired or they merge with stronger companies to avoid failing entirely. This helps a small minority to emerge as the defacto leaders. I believe that cloud companies will survive either through consolidation or standardization, which means cloud companies that have evolved to serve more than one market, which in turn helps drive down costs.

Let me illustrate:

Recently, a report came out that repatriation, or moving some workloads back to on-premise, has resulted in quite a bit of cost savings for companies like Dropbox, Crowdstrike and Zscaler, who use hybrid approaches. The report is quite surprising as the conclusion is that $100 billion to $500 billion in market value is lost on cloud deployments in terms of margins. One use case that is detailed is Dropbox, a company that reported savings of $75 million in two years after repatriation, which in turn, helped the company’s gross margins increase from 33% to 67%.

The problem with cloud is that it’s not uncommon for companies to spend about 60% of their revenue towards committed cloud spend. The solution is aggregating services and products to drive down costs.

Two companies we own that offer standardization are Datadog and SentinelOne. Standardizing means interoperability between various cloud environments and integrated interfaces. This is especially important with multi-cloud or hybrid cloud where companies have more than one environment. This is becoming the new normal to prevent vendor lock-in.

If corporations continue to standardize on Datadog’s platform, then the company will continue to capture market share. Since dealing with multiple cloud vendors quickly becomes cumbersome, there is a natural tendency to standardize in tech, especially with software. Moreover, cloud applications need to communicate, so having everything on one platform can make detecting and resolving issues less complex and costly. The cloud industry is on the cusp of this standardization trend with cloud software vendors, with Datadog leading the way. In this way, Datadog is best positioned to benefit from both the rise in cloud usage and the standardization of cloud software.

SentinelOne is a security position we own. Although the company has many competitors in the EDR space, they extend the acronym to XDR to not only include devices and workstations, but to also include other data points on the network, such as containers and cloud-native applications, and also across the entire stack, such as email, the network, and identity.

SentinelOne uses many data sources to create a data lake. The single pool of raw data is built across a wide range of sources, including other vendors or internal data sources. What matters to customers is that every threat is detected very quickly, and SentinelOne proposes a solution that is able to do both because automation and AI is best done at the data level rather than managing thousands of user endpoints to mitigate attacks.

According to SentinelOne, using their products can produce cost savings can be up to 353% – granted this number is a marketing department, however, the point is that any company increasing ROI in cybersecurity has a real chance of taking market share if their product improves the results. The savings quoted is achieved by reducing the amount of cybersecurity tools a company needs by standardizing endpoint security across more data types. The consolidation in this case saves up to $3 million over a three-year period and the enhanced threat detection saves $671K over three years. Due to automation, $1.2 million can be saved over three years by reducing time and employee hours across the IT team.

Big Data and Analytics/ML – Consumption Model is Here to Stay

There is an oft-quoted statistic that 90% of the world’s data was created in the last two years – and this stat is from 2018. The world produces 44 zetabytes of data across the digital universe as of 2020 and there is expected to be 200+ zetabytes of data in cloud storage by 2025. Each zettabyte has 21 zeroes or is 1,000 bytes to the 7th power. By these estimates, we can expect to see up to 5X growth specifically in data centers. Statista places the number at 181 zetabytes by 2025 up from 64.2 zettabytes in 2020.

In regards to data integration in the cloud, this spans from data lakes, to ETL pipelines, cloud data warehouses and object storage. Data fabrics and data virtualization is key to both hybrid and multi-cloud strategies.

The key thing to know about Big Data, Analytics/ML is that these companies will test financial analysts as they do not bill according to subscriptions like many software companies do. Instead, companies like Snowflake, MongoDB and Confluent bill customers based on consumption. This is a relatively new approach to software billing, which makes it harder to model and forecast near term sales.

As data creation, ingestion and storage soar in the cloud environment, cloud software providers are starting to migrate away from subscription agreements, which are fixed, to a consumption-based pricing model, which are uncapped.

Consumption-based pricing has a few drawbacks. For example, its less predictable than subscription revenue and there isn’t a ‘floor’ on revenue, because if consumption declines then so will sales. However, the flip side is also true, consumption billing does not have a ‘ceiling’ on revenue, so if customer consumption rises, so does sales. This uncapped revenue potential is key to why growth could be quite substantial in this category compared to cloud SaaS peers.

Here is a disclosure from Snowflake in the 10Q:

“Consumption for most customers accelerates from the beginning of their usage to the end of their contract terms and often exceeds their initial capacity commitment amounts. When this occurs, our customers have the option to amend their existing agreement with us to purchase additional capacity or request early renewals”

We want CAGRs that are larger than mobile’s CAGR for 5-10 years. According to industry analysts, the CAGR for machine learning is at 38% between 2022-2029, growing from $15 billion in 2021 to over $200 billion in 2029. Some of this is being driven by Big Tech yet as more companies seek a vendor agnostic approach and multi-cloud workloads, there is ample room for agnostic companies to do well.

Compare this to the smartphone market which grew at 24.9% CAGR with some years in the 12% CAGR range. Here’s an example of a reference for CAGR during Apple’s high-growth years: “The Asia-Pacific smartphones market shipment stood at 87.8 million in 2010 which is expected to reach 294.1 million in 2015 growing at a CAGR of 27.3% during 2010 – 2015.”

Here's how Datadog’s CEO describes what is going on in terms of big data in the Q2 earnings call: “it's almost a given that there will need to be a different way of charging for capturing some of the value provided to customers that can't just be attached to the straight volumes of data that are being exchanged because those volume of data are exploding exponentially while our customers' revenues are not going to explode exponentially.”

To capitalize on the Big Data, Analytics and ML trend – which we fully believe has the potential to produce a FAANG – we hold long-term positions in MongoDB and Snowflake. We are comfortable with the fluctuations of the consumption model, which means some volatility at times, as the consumption model will be tied to higher revenues in the long-term.

Note: We hold a 1% position in Confluent which translates to a lower conviction than MongoDB and Snowflake for this trend. We have recently trimmed 2.5% from Snowflake with the goal of building a bigger position in MDB. Please reference Knox’s trade alerts for more information on these positions and others in real-time.

Snowflake Resources:

MongoDB Resources:

SentinelOne Resources:

Compartmentalizing Cloud

Big Data and Analytics/ML

Confluent

The Blockchain is Eating the Internet

We encourage tech investors to look at cryptocurrencies with a level head. It’s easy to dismiss the blockchain as a fad and it’s also easy to gamble on crypto for a quick gain. We think both approaches are wrong. Instead, our approach is to fully embrace the blockchain and it’s volatility by being willing to trim when the uptrend hits our targets close to the top and layer back in around the bottom.

Knox has a strong track record in navigating Bitcoin’s volatility and we fully expect to continue to trim at the top and layer-in at subsequent bottoms for the next five years – with real-time trade alerts sent to our premium members.

We own the following cryptocurrencies in a longer-term fashion: Bitcoin, Ethereum, Chainlink and Avalanche. The first three come with a higher conviction simply due to the size of their ecosystems yet we think Avalanche stands-out as a secure, decentralized protocol that can scale.

We also own very small, token positions in what we call a YO/LO portfolio (You Only Live Once) where we are a bit more liberated to take higher risks than we would with our core portfolio. Reference the resources below for more information.

Bitcoin:

We covered Bitcoin within a month of launching our premium site in 2019 and it’s in my top 5 for FAANG status. Notably, we had predicted Bitcoin would reach $1 trillion market cap when it was selling off from the $7-$10,000 range to $3,000 range.

The primary reason we are proponents of Bitcoin is that it is the world’s most secure financial network with a higher level of security than the 10,000 global banks combined. This solves a genuine need for the financial system as payments and transfers cannot be automated without a decentralized blockchain solution.

Crypto transfers eliminate processing fees and also hedges against inflation. There are transaction fees charged by crypto exchanges but these fees are not inherent to Bitcoin and will lower in time with more competition.

Apple, Google, Microsoft, and Amazon crossed market caps of $1 trillion because their products scale to global populations and are required on a daily basis. Bitcoin not only scales to global populations, but it also protects their livelihood – a necessity rather than a convenience. This is why we see populations that are not necessarily tech-savvy most enthusiastic about Bitcoin. In 2019, I argued that Bitcoin will reach a $1 trillion market cap as solving a real financial need for global populations should be worth as much as a search engine, enterprise software, social media network, warehouse fulfillment (AMZN), or iPhone hardware company.

In our original report we used the example of Venezuela, where during a period of hyperinflation, the price of a cup of coffee rose to 2,800 bolivars up from 0.75 bolivars within one year, representing an increase of 373,233%, according to Bloomberg data. Essential goods such as toilet paper and medicine were also very costly.

Bitcoin was immediately available to Venezuelans as a store of value and offered them an option to cross the border and escape an autocratic regime. Since then, El Salvador has adopted Bitcoin as their country’s currency.

Currently, the United States is at debt levels of about 133 percent of gross domestic product (GDP). There has been a steady rise in the level of national debt to GDP due to decreased tax revenue and increased spending, especially on health care.

The United States is unlikely to see hyperinflation to the level of Venezuela (at least, let’s hope not). However, trust in fiat currencies will erode as debt continues to climb.

Japan is an excellent case study for an economy that is struggling due to quantitative easing. The Japanese debt-to-GDP ratio is at an all-time high at 254% due to its quantitative easing. Government debt to GDP in Japan averaged 137.4% from 1980 to 2017.

Easy money policies from Japan’s central bank harmed domestic asset returns by suppressing local interest rates. Ranking as the world’s third largest economy, Japan resorted to negative interest rates in 2016. In April 2016, it was reported that a “Japanese bank buying 5-Year U.S. Treasuries with perfectly hedged currency and duration risk would (lose) 0.9% a year.”

Consequently, Japan is a thriving bitcoin market and has seen increased crypto deposits. According to the Japan Virtual and Crypto assets Exchange Association (JVCEA) Japan’s virtual currency deposits recorded 1.41 trillion yen or about US$13 billion in March last year, the volume was about seven times more than in March 2020.

During the recent Ukraine-Russian war the use of crypto has once again taken prominence. The Ukrainian government has also accepted crypto donations during this crisis. In the words of Alex Bornyakov, Deputy Minister of Ukraine’s Ministry of Digital Transformation, “In times like these, response time is crucial. Crypto is playing a role to give us flexibility to respond really quickly to deliver the army’s required supplies.” The lack of financial access might also increase the use of crypto in both the countries. The Ukraine central bank had suspending electronic transfers and reduced cash withdrawals and there were reports that Ukrainians were turning to cryptocurrency.

According to Alex Gladstein, Chief Strategy Officer at the Human Rights Foundation, “The fact that it can’t be frozen, the fact that it can’t be censored, and the fact that it can be used without ID is very, very important,” He further added, “And they are why bitcoin is such an important humanitarian tool.”

We’ve written extensively on Bitcoin and we encourage you to read more about the importance of the Lightning Network in our resources below, which is a payment protocol that operates on top of cryptocurrency blockchains and enables fast transactions.

The Lightening Network will be used for small transactions that don’t require the security of the bitcoin network. Large transfers that require decentralized security will continue to take place on the original layer.

The final iteration for the Lightning Network will be the cross-chain atomic swaps, which will exchange crypto tokens between different blockchains without the need for a crypto currency exchange.

Benefits of the Lightning Network:

Transactions will take place on the Lightning Network channels and outside of the blockchain:
Fees will be minimal to non-existent for small payments like coffee, dinner, and local stores.
Quick transactions no matter how busy the network is. The transactions will be instantaneous and able to keep pace with Visa, MasterCard and Paypal.
Cross-chain atomic swaps will eliminate the need for separate crypto exchanges.
The Lightning Network can reach 1 million transactions per second.

Bitcoin Resources:

Layer One Networks

If you want a perfect parallel to the mobile duopoly of Android and iOS, then it will be Web3. We began with artificial intelligence because by increasing GDP, AI/ML promises to be the technology that delivers the most gains in the public market’s history – far exceeding mobile. Yet, the blockchain offers a parallel to mobile as what layer one networks set out to achieve is very similar to what Apple’s app store achieved.

The primary difference between Ethereum and Bitcoin is that Ethereum is not trying to compete as a currency. The focus of Ethereum is on its network, not the coin. Butkin’s vision is to create an open network for decentralized applications (dapps) and smart contracts based on the Turing complete programming language Solidity. The takeaway is that just like Apple hosted apps on its operating system, Ethereum hosts d’apps on its network.

These three benefits are: decentralization, security and scalability. The issue is that most decentralized networks cannot offer all three without some compromise.

Ethereum faces constraints in transactions per second (TPS) and how to overcome the high energy costs of mining that comes with decentralized security. The network simply can’t scale without the upcoming release of Ethereum 2.0.

In our premium analysis last year on Ethereum here, we discussed the difference between Proof of Work (PoW) and Proof of Stake (PoS). In addition to the Proof of Stake merge that Ethereum must complete, the network must also launch shards. Nodes in the previous network must download a transaction, calculate it, archive it and read every transaction in Ethereum’s history, which is terribly inefficient. Shards create a subset of the network where nodes are dispersed for more efficient processing. Ethereum 2.0 must also replace Plasma with ZK Rollups, which allow for hundreds of transfers to be rolled into a single transaction.

In November, we wrote another update on crypto and Ethereum, stating that the expectation was for Proof of Stake to merge in late 2021 with Shards and Rollups expected by late 2022 or early 2023. The timeline for PoS is delayed yet again until Q3 2022, which puts Sharding and Rollups out another year potentially to Q3 2023. (Read more in the resources below).

The takeaway: Ethereum has a wide lead in terms of number of d’apps and developers (remember that developers adopting CUDA created Nvidia’s moat). However, the Merge to Proof of Stake is an unknown which leaves the Layer One network market wide open for now.

Avalanche

Layer One Networks are considered early-stage tech investing which carries higher risk. Ethereum clearly has a head start, and after the proof of stake merge, we could see the network take off in a meaningful way.

However, there are other Layer One networks to consider. We hold an allocation in Avalanche due to it’s Ethereum Bridge, application-specific subnets, and the launch of a consumer-facing app over the next quarter. Avalanche also has a high Nakamoto Coefficient, which is a number that designates the number of nodes that would need to be corrupted to slow or prevent a chain from functioning properly.

Avalanche launched with three chains. Per our YO/LO write-up: The X-Chain is for creating and exchanging assets including NFTs, the P-Chain validates and creates subnets, and the C-Chain is for executing Ethereum Virtual Machine (EVM) contracts. The C-Chain offers interoperability with Ethereum, which is why the Avalanche bridge is the most popular ETH Bridge currently. The P-Chain is what is used to create and manage subnets. The coordination of Avalanche validators occurs on the P-Chain and it can support thousands of subnets and millions of validators.

As we stated in the AVAX write-up: “Ethereum is running into issues with 500,000 to 1 million daily active users. Meanwhile, a single mobile application sees hundreds of millions of users, such a Twitter or Spotify. What Layer 1 can handle this level of adoption? That is a platinum-level question for investors to answer. To be clear, it could be Ethereum in 2023 if the developers and users prefer to not migrate. However, if the ecosystem runs out of patience and seriously looks for an alternative, then Avalanche is a candidate.”

Ethereum Resources Here:

Avalanche Resources Here

Chainlink:

Warren Buffet famously said: “The stock market is a device for transferring money from the impatient to the patient.” I believe Chainlink could be our best performing asset in our portfolio by 2030 as the middleware that enables smart contract through decentralized oracles.

Smart contracts are a more advanced use of blockchain where an exchange between two parties is automated based on conditional provisions. These self-executing contracts are written into lines of code, and the agreements contained exist across a distributed, decentralized blockchain network.

Smart contracts offer a more complete use for blockchain. First discussed in 1996 by Nick Szabo, some claim that smart contracts are the real use case for blockchain as they aim to automate financial transactions, and in the future, can automate machines.

We have written extensive deep dives and webinars on what the company does but for those who would rather get the elevator pitch, it’s this: Chainlink is the most likely candidate to become the Google of Web3. In fact, ex-Google executives are joining Chainlink as strategic advisor, former CEO Eric Schmidt, new Chainlink Chief Product Officer, Tensorflow’s Kemal El Moujahid.

We are very bullish on Chainlink and it was our first one-hour deep dive webinar for this reason. However, it requires a longer-term mindset, which we certainly have at the I/O Fund. Our goal for our position is sizable gains by 2025 with an exit in 2028-2030.

Chainlink Resources Here

FAANG Isn’t Dead

“Winners keep winning” is a reliable and true statement. We began this analysis by showing you a chart of how the world’s most valuable companies change every 10 years. However, there is an important caveat: tech overtook oil to become the world’s most valuable industry in 2010 and we have yet to see the pattern that tech sets in terms of how often the top 5 will rotate now that tech is in the driver’s seat.

Microsoft:

We were one of the first analysts to cover Microsoft Azure’s hybrid computing strategy and why that could narrow AWS’ lead in the cloud IaaS market. At the time, tech was selling off in Q4 2018 yet we were firm that Microsoft would emerge as a significant leader in this space by specializing in a mix of on-premise and cloud deployments.

Hybrid cloud allows for scenarios where customers can keep their most sensitive data on their own servers while sending workloads to the private or public cloud that gain an advantage from mining data more efficiently and requires improved accuracy and productivity.

Azure’s strength in offering both on-premise and cloud in a hybrid solution has prompted Amazon to chase Microsoft with recent efforts to improve its hybrid strategy. Today, Azure claims more than 95% of the Fortune 500 as customers because of its hybrid flexibility.

Azure growth of 46% is performing quite well given the tough comps it has overcome, and Microsoft’s best financial metric during this tech selloff is that commercial bookings increased 28% this quarter following 32% increase last quarter. I would look for Azure to remain elevated against AWS and Google Cloud for those two reasons – hybrid cloud leader which attracts large enterprises and its ability to reduce costs with its tech stack.

In our latest Q2 webinar, I discussed why reducing cloud costs is a key trend for 2022 and beyond. To put it simply, Sayta Nadella said in this quarter’s earnings call: “More value for less price means you win.” In the same breath, he also said: “Most businesses are not looking to their IT budgets or to digital transformation for budget cuts.” These two statements echo my first point in the Q2 2022 webinar which is that both are true: increase in cloud spending will continue and companies will want to lower costs associated with cloud.

That’s going to be the trick moving forward – which companies assist cloud migrations while lowering costs. Microsoft is the leader here as the company aggregates many cloud services and products under one umbrella which creates substantial leverage to undercut on price.

Microsoft is increasingly becoming a cybersecurity company, as well, with $15 billion in revenue and growing at a rate of 45%. Microsoft was careful to build a multi-cloud product and is the only Big 3 cloud vendor to be multi-cloud on security right now. This also helps to drive down costs for Microsoft’s customers.

There are additional catalysts for Microsoft beyond Azure’s winning streak, its large security footprint, and the ability to lower costs. The first catalyst is that Microsoft is setting up to own the edge through its telecom partnerships. Another catalyst is that when more enterprises adopt AI/ML, whether it’s automation, super computers and/or other use cases for training and inference, it will a natural decision to use Microsoft if they’re already optimized for Azure. As discussed, enterprises will drive forward the next major market in tech (AI/ML) and Tier 1/Fortune 500 will be the largest customers for AI/ML. Power Automate was up 72% year-over-year, surpassing $2 billion in revenue, and this is only the beginning.

Third, the company ranks with Nvidia and Unity for inroads to the Metaverse as it owns many gaming publishers now and is the most widely used VR headset (HoloLens). The company also has Teams to introduce Metaverse-like qualities to business meetings. It will be industrial that drives forward 3D worlds (not consumer) and Microsoft is auspiciously positioned.

Microsoft Resources Here:

Alphabet:

We have been meticulously building our portfolio for the next FAANGs of 2030 since we launched the site in 2019 with the understanding that even getting 1 or 2 correct can create generational wealth. Our goal from the beginning has been to stick with our winners and to cut our losers and we have compiled quite a bit of research along the way.

Alphabet is a new addition to our portfolio and one we’ve been circling for some time. If the 2014-2022 era in digital advertising was known as the walled garden era primarily fueled by third-party data then 2022-2030 will be known as the brick walls of first-party data – meaning, publishers controlling their data become the trend that drives forward digital advertising as we move into more AI/ML-driven ads.

In fact, we are in the midst of what is the biggest shift in digital advertising since advertising went digital. The shift is due to Apple and other real estate owners shutting off how data is collected across mobile and desktop. In the mobile era, third-party data was rampant but all of that changed with the release of iOS 15.

Note: We were one of the top authors on this topic with coverage dating back two years before the change occurred on both Forbes and MarketWatch here.

Google will be following in Apple’s footsteps by changing how third-party data is collected on Android and Chrome. This will greatly strengthen the company as not only is Google an equal or greater real estate owner compared to Apple but the company is also a publisher for the purpose of ads with Search and YouTube. This means long-term ads placed on Google will be more effective and produce higher ROI than those with less signals to work with.

These data collection changes are coming just in time for the advantage from first-party data to be realized across AI/ML (with digital ads) and a myriad of other uses cases.

Google Resources Here:

Consumer Isn’t Dead

We’ve focused quite a bit on enterprise for the purpose of this article yet we want to acknowledge that consumer-facing tech carries strength in most macro environments.

We hold the following stocks to capture consumer-driven gains. Notably, we’ve covered in the past how supply chains are contributing to consumer spending and inflation. We are watching this closely for when growth in this area rebounds. You can access our research on this here.

Roku: Netflix made it into FAANG and CTV ads will be a bigger market than subscriptions – primarily CTV ads will do well globally. Roku must prove its hardware strategy will pay off in global markets starting with LatAm.
Snap:: This company has had a brutal month – yet audience metrics have been strong post-Covid and the Gen-Z/Millennial concentration is important to take note of.
Shopify: This company is spending an unknown amount on the fulfillment center yet can rival Amazon simply through unlimited distribution channels. Social commerce will eventually take off despite the setback from Apple’s iOS changes.
Twilio The Twilio management team is known to be visionaries and they are pivoting into an API-forward marketing platform with strong PII data – they are early to API-driven automation taking over marketing and advertising.

Conclusion:

Thank you for taking the time to read this report. Despite tech being one of the most volatile sectors in the market, it is also the most rewarding. Had you entered Apple between 2008-2010, it would have blown away all other positions in your portfolio across all other industries. The same goes for a Facebook or a Google position. Half the battle is finding them, which we think we are particularly well suited for, and the other half of the battle is knowing which stocks to sell and which ones to hold when the tide rolls out. We show you how we do this with real-time trade notifications plus a hedge for ample insurance during drawdowns. There are many positions we own today that we will own in 2030 with the goal of perfectly timed exit. We are patient and thorough in our research as we acknowledge and accept that approximately 5 companies will lead to 90% of our wealth in 10 years.

SentinelOne: Q1 Earnings Review

Posted on June 3, 2022June 30, 2026 by io-fund

SentinelOne continues to be the strongest cloud stock on the top line. This earnings report did not disappoint on the top line with 109% revenue year-over-year of $78.3 million compared to consensus of $74.64 million. ARR was up 110% year-over-year to $339 million. Customer count with ARR over $100K also outpaced revenue growth at 131% increase. Net retention rate grew to 131% which is above the 130 line.

The company is expecting growth of 109% at the midpoint for next quarter with $8 million coming from the Attivo acquisition. My notes have analysts expecting $84.83 million so if we add the $8 million for $92.83 million, the company is beating those estimates (consensus would have been for 103% growth including Attivo). Organic is expected to be in “the low to mid 90%” range, which reflects a raise from the 85% consensus for Q2.

The company raised full year organic revenue to “the midpoint of 80% range” up from the previous guidance of 80%. The full year guidance was raised to 98% including Attivo for revenue of $405 million. The company stated Attivo would contribute $30 million to full year revenue (although one analyst felt the math was off and has Attivo contributing $35 million). Previously, organic revenue growth was guided at $368 million for FY2022 with analyst consensus slightly higher at $370 million.

To SentinelOne’s credit, the company offers clear communication about margins. It’s one of the few companies where the CEO will discuss this at the onset of the opening remarks.

Per the analysis on compartmentalizing cloud stocks here, we went into the earnings report wanting to see an improvement in operating margin. The company was expected to report (86%) to (84%) Non-GAAP operating margin and provided a beat at (73%). This is up from (127%) last year for an expansion of 54%.

The GAAP operating margin remains an eyesore due to stock-based compensation at (115%) of revenue, up from (165%) in the year-ago quarter.

SentinelOne demonstrated strong improvement in gross margin from 51% in the year-ago quarter to 65% in this quarter. It’s up 2 basis points sequentially and up 15% YoY and is the highest GM for the preceding four quarters.

The one thing that could have weighed on the AH price action was the guide on operating margin for Q2 being (75%) to (73%) – as the critical point here for SentinelOne is that the full year guide management has provided for two quarters is for Non-GAAP op margin for FY2023 to be (60%) to (55%). Even though Q1 was a beat on Non-GAAP margin, the path to delivering on the guidance becomes a bit obscure the longer we remain above this level.

The reason we’ve accepted the weaker (albeit improving) margins is that the company is working towards being FCF positive by 2025 and is not likely to raise cash before this occurs. Any change to this would cause us to look at our thesis again.

The company expects to improve adjusted gross margin to 69%-70% and if we assume similar GAAP percentage as this quarter, it would be about 66% GAAP GM for FY2023.

Singularity Cloud was the company’s fastest growing module, growing over 50% sequentially.

Management focused on the strength of their MITRE Attack results with 100% protection, 100% detection, 100% real-time protection, 99% visibility, 99% analytic coverage. I’m sure we will hear Crowdstrike’s response to this on Thursday 🙂

One analyst asked about the European segment and management stated there is a wholesale movement away from Kaspersky either by choice or by mandate and this is a tailwind for them. Secondly, the Broadcom-VMWare acquisition is favorable for them as they are now capturing CarbonBlack business. In terms of taking business from these two vendors: “We expect that to accelerate in the quarters to come.” I’ll expand more on this when I get the transcript.

Conclusion:

We had said the following in our cloud update:

“Will SentinelOne be able to provide a meet/beat on operating margin in the upcoming quarter and a meet/beat for the full year guide? This must happen and we also need revenue to remain strong.”

SentinelOne’s operating expenses were front-end weighted last year with Q1 being the steepest operating loss and the year getting progressively better (nearly 100% improvement on weak numbers).

If last year is any guide, then SentinelOne is capable of meeting their full year guidance of (60%) to (55%). The company did beat its operating margin guidance this quarter and revenue was strong including key metrics.

I continue to believe the key to this stock is the ongoing revenue strength and its ability to prove to analysts and institutions that it’ll generate cash by 2025. Due to Crowdstrike being a close comparable, it’s likely SentinelOne can (and must) follow in Crowdstrike's cash efficient footsteps, which is what the market will want to see. The product continues to prove itself as exceptional and there was evidence of this in terms of high ARR customer growth, beat/raise on revenue, strong growth on cloud product, and we are likely to see nice movement in the identity product soon. We are keen on SentinelOne's ability to standardize multiple areas of cybersecurity and to do so at a high MITRE ranking.

Despite the red AH on the stock, I see no notable issues with this ER from my perspective.

Previous product analysis is located here:

SentinelOne Exceptional Product at a Decent Valuation

SentinelOne: Strong Defender and Q4 Review

Semiconductor Stocks: Q2 2022 Overview

Posted on June 3, 2022June 30, 2026 by io-fund

This article was originally published on Forbes on May 28, 2022,11:54pm EDTForbes on May 28, 2022,11:54pm EDT

Semiconductor stocks have gained prominence due to growth drivers such as artificial intelligence, high-performance computing, 5G, robotics, machine learning, and electric vehicles. Despite supply constraints and the challenging macro environment, semiconductor stocks have withstood the tech sell-off better than other sectors. This is due to many semiconductor companies being profitable with strong free cash flows.

We reviewed the stocks in the sector to find out which companies stand out in terms of revenue growth, profits, cash flows, and earnings surprise.

Top 20 semiconductor stocks with highest growth rates for the current fiscal year.

Source: YCharts

In the above chart, Indie Semiconductor leads with the expected growth of 130% year-over-year in the current fiscal year. The company is riding the growth trend in advanced-driver assistance systems (ADAS) and electric vehicles. It has a Serviceable Addressable Market (SAM) of $40 billion by 2026. The company supplies chips and software to the automobile sector. It’s chips power sensor capabilities like LiDAR and Radar, and vehicle electrification.

The company’s revenues accelerated by 171% YoY to $22 million in the recent quarter. The management expects revenue to grow 178% at the mid-point in the next quarter. While the company is not profitable at the moment. The management expects it to be profitable in the second half of next year.

AMD is expected to grow 60% this year due to the Xilinx acquisition. The company had initially guided for organic growth of 31% during Q4 results. The Xilinx acquisition was completed in February this year, and partly by better demand from end markets. In the recent quarter, the company’s revenue grew by 71% YoY to $5.9 billion, with organic revenue growth of 55%. Even if we exclude Xilinx, the company is a leading growth stock among the semis due to data center growth and gaming.

Top 20 semiconductor stocks with highest growth rates for the next fiscal year.

Source: YCharts

Navitas Semiconductor has the highest growth rate in the above chart. The company is a leading player in the Gallium Nitride (GaN) chips. The benefits of GaN include fast charging and better power efficiency. Currently used in mobile phones & laptops, EVs are the future opportunity. Ambarella is another interesting company to watch. The company’s chips which were previously popular for using in drones and cameras have recently found a niche in the automobile sector. The company’s AI computer vision chips benefit from the Internet of Things, ADAS, and autonomous driving. The company’s revenue in the 4Q FY2022 grew by 45% YoY to $62.1 million. The computer vision revenue accounted for more than 25% of the FY2022 revenue and is expected to be 45% of FY2023 revenues.

Semiconductors with Top Forward P/S Sales multiples

Source: YCharts

In the above chart, SiTime Corporation has the highest forward P/S ratio. The company is a leading provider of Silicon Timing Solutions. In the recent quarter results, the company’s revenue grew by 98% YoY to $70.3 million. The revenue is expected to grow 50% this year and 23% in the next year. The strong growth rates are reflected in the company’s share price, which has doubled in the past year.

Sign up for I/O Fund's free newsletter with gains of up to 403% – Click hereSign up for I/O Fund's free newsletter with gains of up to 403% – Click hereClick here

Wolfspeed is another leading company that has a premium valuation due to the company’s expertise in Silicon Carbide chips. The company’s revenue is expected to grow 38% this year and 44% in the next year. According to MarketsandMarkets, the Silicon Carbide market is expected to grow at a compound annual growth rate of 19% from 2021 to 2026. Hybrid and electric cars are the future growth drivers for Silicon Carbide.

The company recently entered a deal with Lucid Motors to supply Silicon Carbide devices from the newly opened Mohawk Valley Fab. According to Gregg Lowe, CEO of Wolfspeed, “As the world advances towards an all-electric future for transportation, Silicon Carbide technology is at the forefront of the industry’s transition to EVs, enabling superior performance, range and charge time. Our investment in the Mohawk Valley Fab ensures our customers, including Lucid, have access to the advanced products they need to deliver innovative solutions to the market.”

Nvidia has seen some weakness recently due to the broader tech sell-off. However, the company deserves a premium valuation due to the company’s growth prospects in the AI data center and solid long-term prospects in the automotive chip industry.

Quarterly Revenue Surprise

Source: YCharts

Cirrus Logic crushed analysts’ consensus revenue estimates by 17%. The company’s Q4 FY2022 revenue grew by 67% YoY to $490 million. The company’s guidance for the next quarter is between $350 million to $390 million, representing a YoY growth of 26% at the mid-point of the guidance. It was higher than the analysts’ consensus estimates of $295 million. John Forsyth, CEO of the company, said, “We delivered strong financial results in FY22 as revenue increased 30 percent year over year driven by high-performance mixed-signal content gains.”

Top 5 ranked semiconductor stocks based on Free Cash Flow Margin

Source: YCharts

Cirrus Logic not only beat analysts’ revenue estimates it also ranked the highest among the semiconductor companies with the highest free cash flow margins. This is an important financial metric in the current environment as we have noticed in the current earnings season that many companies that fell short in this metric the shares got sold off.

Top 5 ranked semiconductor stocks based on Net Profit Margin

Source: YCharts

In the above chart, Indie Semiconductor, which we discussed earlier in our article, also ranked the highest among the companies with the highest net profit margins. Intel ranks third in the category. However, the company faces significant competition from AMD, which can be seen in the lower valuation the company is trading.

Micron Q2 FY2022 Update

Posted on April 5, 2022June 30, 2026 by io-fund

Micron reported Q2 results that beat estimates and guided for Q3 sales and earnings growth above expectations. However, Micron’s share are off 9% post-earnings, which may be due to commentary around soft demand in China, slowing PC sales and concerns that Micron may be nearing the top of the cycle. We believe that these concerns are temporary, and that Micron is structurally becoming a less cyclical company, which deserves a premium multiple. I discuss the company’s latest results and why we believe the recent sell-off is overdone in more detail below.

Micron’s Q2 FY2022 Beat Expectations and Guidance Was Above Consensus

Micron reported Q2 FY2022 results on March 30th, and sales increased 1% QoQ to $7.8 billion driven by a 4% sequential rise in NAND sales, which accounted for ~25% of total revenues. NAND sales also increased 19% YoY and management expects NAND sales to increase by ~30% YoY for the year. Demand for NAND is being driven by Micron’s new 176-layer NAND technology, which represented the majority of Micron’s NAND shipments. We explained the importance of 176-layer NAND here, stating that Micron has significantly increased memory capacity and is a leader in this technology, allowing the company to capture more market share.

Importantly, the strength in NAND should also be a tailwind for Lam Research, which sells the etching equipment necessary to build the layers for 176-layer NAND. In our latest update on Lam Research, we explained that “the key reason we think Lam could fare better than its peers is because as 3D layers increase, capital intensity also increases. The process does not scale linearly, instead it’s non-linear because it takes longer than 2X to etch a stack that is 2X high and requires more complex etch and deposition equipment”. With Micron guiding for $12 billion in Capex this year and plans for $150 billion in capacity expansions over time, we should expect Lam Research to see strong demand for etching equipment going forward.

Micron’s NAND prices also benefitted from the contamination of ~8% of the global supply of NAND. In February, memory peer Western Digital disclosed that there was a contamination event at two of its Japanese JV facilities, which resulted in 6.5 exabytes of NAND memory being contaminated. Likely benefitting from this event, Micron’s Q2 NAND prices rose ~4% QoQ, driving much of the topline growth as volumes were flat. As shown below, Micron has outperformed relative to Western Digital YTD, however, both companies have underperformed the broader market in 2022. I outline a few reasons for this in more detail below, which we believe are only temporary.

Similar to NAND, DRAM sales increased 2% QoQ and were up 29% YoY to $6 billion, or 73% of total sales. DRAM volumes increased but were offset with a decline in ASPs. Strong demand in datacenter drove the increase in DRAM sales. For example, Micron’s largest segment, Compute and Networking, grew sales 31% YoY to $4 billion, driven by a 60% YoY rise in data center sales which were “supported by robust demand across our DRAM and SSD portfolio” (Q2 call, 03/30/22). DRAM sales benefitted from Micron’s leading 1-alpha technology which is increasingly being adopted in the memory-intensive cloud environment. During the Q2 call, CEO Sanjay Mehrotra stated that DRAM sales will continue to ramp into 2023 when he said that “We have broadened the qualifications for our 1-alpha DRAM products and are well positioned to support the data center DDR5 transition driven by new CPU platforms, which are targeted to begin ramping later this calendar year and gain momentum in 2023”.

Following the strength in cloud sales, Storage sales increased 38% YoY to $1 billion as SSDs continue to replace HDDs, while Embedded sales increased 37% YoY driven by strength in automotive. A blemish was weakness in Mobile sales, which increased just 4% YoY to $1.9 billion. While the rollout of 5G phones will lead to a ramp in memory content per phone, there may be demand headwinds on the horizon. For instance, Apple cut its forecast of 5G iPhone shipments by ~20%. I discuss this in more detail below.

Continuing down the income statement, gross margin increased by 2,100 bps YoY and 100 bps QoQ to 47%, benefitting from higher NAND margins and the ramp in 1-alpha DRAM and 176-layer NAND technologies, which reduces costs as it scales. Management noted on the Q2 call that most of the efficiency benefits have been realized, and that margin expansion from the ramp is largely behind the firm. Furthermore, YoY gross margin comps were impacted by a one-time $300 million charge taken last year when Micron switched to FIFO accounting.

The strength in gross margin flowed down to operating profit, which increased 118% YoY to $2.7 billion. The dramatic rise in profitability was driven by higher selling prices and cost reductions from the ramp in new technologies outlined above. However, Micron has historically been a cyclical industry, and there may be concerns that Micron is nearing the top of the cycle. This may explain the recent sell-off in Micron’s sales, yet we believe that Micron is becoming structurally less cyclical and that its multiple will rebound once this is clearly evident in future results (discussed in more detail below).

Finally, GAAP earnings per share were $2.00 while non-GAAP EPS was $2.14, which beat estimates by $0.16. Non-GAAP EPS increased 118% YoY and the strength in EPS growth should continue going forward. For instance, Micron will benefit from a lower tax as Idaho’s governor signed a new tax law on March 16th, 2022 that will reduce Micron’s taxable income (Micron is HQ in Idaho). The CHIPS act may also be a tailwind to earnings as the US government looks to incentivize reshoring of manufacturing capacity.

As of the end of the quarter, Micron had $12 billion in cash and equivalents and free cash flow was over $1 billion during the quarter. Management stated that they expect free cash flow generation will be “substantially higher” over the next two quarters relative to H1 2022. Micron intends to use ~50% of its free cash flow to buy back its stock and pay dividends to shareholders. Since 2019, Micron has reduced its share count by an aggregate 113 million, or by 9%. With Micron guiding for record sales and profits in FY2022, cashflow generation should be significant, which will support more buybacks in the future.

Looking forward, management expects Q3 sales to grow 18% YoY to $8.7 billion, which beat initial topline estimates by 6%. Management stated that they are “tracking ahead” of their initial guide set in Q1 for FY2022 and that demand remains strong, but noted during the Q2 call that “there are some pockets where semiconductor shortages have not improved as fast as we had expected, and these shortages are likely to continue into calendar year 2023”. Nonetheless, Q3 adjusted EPS is expected to grow 14% QoQ to $2.46, which beat initial estimates by 9%.

Potential Risks are only temporary

As discussed above, Micron reported strong top and bottom-line results, guided above consensus and expects to be report record sales and earnings in FY2022. However, despite this, Micron has underperformed in 2022 and is off ~9% since announcing FQ2 results. This may be due to a couple developments: 1) softness in mobile and PC sales and in China, 2) and concerns that Micron may be nearing the top of the cycle.

In regards to the first point, during the Q2 call management stated that “We see some weakness in the China market as the local economy slows, smartphone market share shifts and some customers take a more prudent approach to inventory management.” CEO Mehrotra added that he expects PC unit growth will be “flattish”. These comments may have contributed to a post-ER sell-off, and it is notable that AMD is also off following the Micron Q2 print, likely due to its exposure to PC sales. However, management added further color that enterprise PC sales are expected to be strong in the near term, which are more content rich in terms of DRAM and NAND content, which should offset this pressure.

Moreover, 5G phone sales are just now starting to ramp, but the timing of this ramp remains unknown. As mentioned above, Apple has reportedly cut its production forecasts for its first 5G phone by ~20%. While this may be a near-term headwind, it is inconsequential in the long term. This is because 5G phones will inevitably take share from 4G going forward, and 5G phone DRAM content is 50% higher than 4G, while NAND content is >100%. We expect mobile will be a tailwind going forward, despite the near term uncertainty in the pace of the ramp.

Finally, a trend that is typical with highly cyclical companies is that investors tend to reduce exposure when earnings are high due to concerns that the company may be nearing the top of the cycle. Historically, Micron has been a highly cyclical company with periods of oversupply and rapidly declining prices. However, with more demand drivers coming from data centers/cloud and automotive, memory demand is no longer dependent on the short-cycle PC market.

During the Q2 call, CEO Mehrotra explained that over 75% of its quarterly volume are under long-term agreements (LTA) that go out beyond four quarters or more, up from less than 25% in prior years. CEO Mehrotra added that all of the company’s large customers are now under LTAs, which helps improve demand visibility and reduces uncertainty. An increase in LTAs significantly reduces the cyclicality of Micron’s business.

Moreover, new trends on the horizon further smooth demand for memory, reducing Micron’s dependence on the short-cycle PC market. For example, CEO Mehrotra stated that “new EVs are becoming like data center on wheels, and we expect over 100 new EV models to launch worldwide in this calendar year alone”. The memory content in higher end EVs is 15x higher than the average car, which further reduces the cyclical nature of Micron’s business.

As shown below, Micron trades at a 9x PE multiple, which is below where it was trading in 2017 and well below its multiple in 2020 and 2021. We believe that the market remains in a “wait and see” mode until Micron can prove that it is less cyclical. If Micron can prove that it is less cyclical going forward, we should expect a re-rating of its multiple going forward. A trend that supports this is the reduction in finished goods, which declined QoQ despite the softness in China, PC and mobile. A build in finished goods inventory would signal that demand may be weakening, a trend we have yet to observe in the memory market.

SentinelOne: A Strong Defender and Q4 Review

Posted on March 28, 2022June 30, 2026 by io-fund

SentinelOne: A Strong Defender and Q4 Review

The marketing language between Crowdstrike and SentinelOne is thick, and I don’t expect this to change anytime soon. We’ve called SentinelOne a “stellar product” in our first analysis and an “exceptional product” in our second analysis. This is based not only on detection, where Crowdstrikes rank high, but also on accuracy of triggering a response.

The product differentiation is best summed up by the fact other vendors require data to be sent to the cloud for analysis and often have many humans monitoring the alerts to take action. Meanwhile, SentinelOne uses automation to find the threat which reduces the number of false positives. Instead of getting every piece of telemetry that requires the security team to investigate, SentinelOne’s endpoint detection and response solution eliminates the noise so that the security team is only responding to those that have the potential to be critical.

SentinelOne often emphasizes the fact that legacy antivirus powered by human-generated signatures still remains a widely used security technology. This is in spite of the fact that they are ineffective and reactive. Human-powered endpoint detection and response, or EDR, emerged as the alternative in which people became the detection and response crew.

Speed is everything in security and SentinelOne asserts Crowdstrike’s 1-10-60 rule is outdated. The 1-10-60 response rule claims the best achievable cybersecurity outcome is capped at one minute to detect an attack, 10 minutes to investigate, and 60 minutes to respond. Meanwhile, recent ransomware attacks have proved that it only takes milliseconds to breach an organization and cause damage.

Data + AI = Storylines

SentinelOne believes that cybersecurity is fundamentally a data problem. The company’s Singularity Platform ingests, correlates, and queries petabytes of structured and unstructured data from ever-expanding disparate external and internal sources in real-time. It builds rich context and delivers greater visibility by constructing a dynamic representation of data across an organization. As a result, the company’s AI models are often highly accurate in triggering a response.

The company’s distributed AI models run both locally on every endpoint and every cloud workload, as well as on the company’s cloud platform and the AI models predict threats in milliseconds. The behavioral AI model maps and links all behaviors on the endpoint to create Storylines. When an activity is deemed to be a threat, the system automatically takes action to kill the attack.

In the cloud, the company’s platform aggregates these Storylines. The streaming AI detects anomalies that surface when multiple data feeds are correlated with additional external and internal data. By providing full visibility into the Storyline of every secured device across the organization through one console, the platform makes it fast for analysts to search and hunt for threats.

Storylines maintain a complete record of unauthorized changes that are made during an attack and can roll back or remediate any damage done during an attack very quickly. The ability to turn back time on a device is unique to SentinelOne. According to the company, this is the “ultimate safety net” and eliminates costly incident cleanup.

S versus CRWD = No Clear Answers

The challenge for analysts and investors is that Crowdstrike also has a strong product and is certainly the heavyweight in the space with SentinelOne being the defender. We’ve written a more thorough overview on product comparing SentinelOne and Crowdstrike here.

In our previous analysis, we’ve discussed SentinelOne’s strength in ranking at the top in Peer Reviews despite ranking lower across industry analyst reviews. The most recent MITRE ATT&CK evaluations showed a very strong report for Crowdstrike on 100% detection yet showed very strong results for SentinelOne on 100% visibility and no misdetection.

If public investors are looking for clarity directly from the source, there will be none offered as these two companies are committed to battling it out on earnings calls, marketing materials and anywhere else they can find an opening.

It probably says something about S, however, that CRWD would even bother to defend itself given its annual revenue will be over $2 billion when S is at roughly $400 million. In other words, why bother with a competitor 1/5th your size in a crowded endpoint security vendor market? I imagine it’s because S makes bold, ongoing statements like this in their earnings calls and elsewhere: “We've maintained incredibly strong win rates in all competitive situations against both legacy and next-gen vendors. This is because of the breadth of our platform, covering endpoints and surfaces of all types, cloud workloads, Kubernetes, mobile devices and IoT devices and soon, identity.”

Cloud Security Segment Could Rival Endpoint Security

Cloud is a growth lever for SentinelOne as the company leverages a microservices architecture for rapid and frequent updates. The company offers support for Kubernetes workloads with additional runtime protection and simplified deployment.

In our previous write-up, we had stated this was the most important statement on the Q3 call:

“Cloud still remains our fastest-growing module. About 10% of endpoints are covered by cloud and servers. It has been our fastest-growing module for some time. Cloud is a piece of the business, I think that we think will expand greatly in the future. We anticipate that at some point, it will be the similar size to the endpoint market.”

The company stated that across the company, it’s the “cloud workload protection and data retention modules that outperformed the most with year-over-year ACV growth of over 10X.” We are talking outsized growth on small numbers but this is key to watch moving forward with the more revelatory statement long-term coming from Q3 about the market matching the endpoint market.

In August, the company released SentinelOne Storyline Active Response (STAR) which is the cloud-based automation engine that allows security teams to create custom detection and response rules.

The company stated DataSet was up 20% year-over-year. The product DataSet is what became of last year’s Scalyr acquisition, a leading cloud-native data analytics platform that serves as a big data engine for the XDR platform. This speeds up the process and drives down costs by ingesting and correlating terabytes of data at machine speed and also makes SentinelOne more competitive against SIEM tactics for data correlation and response.

Here is the advantage of DataSet when comparing to the competition, as stated in the call: “And I think that a lot of the large enterprises out there are looking at that because otherwise, if they're going with a different EDR vendor, they suddenly need to figure how are they going to retain data on the back of maybe a different platform. And that's sometimes going to be just incredibly costly. With us, it comes on the back of a single platform.”

Attivo Acquisition

SentinelOne extends the definition of XDR beyond “endpoint’ to include other data points on a network, such as containers, cloud-native applications, and even email or messaging. With the Attivo acquisition, the company will have more in-network and identity detection in the event a hacker is mimicking an employee or authorized user. This is especially important given the recent SolarWinds hack.

Attivo Networks specializes in Active Directory, which helps IT departments connect users to Windows-based IT systems. This means it could help SentinelOne’s cross-sell into more traditional enterprises, which was alluded to on the call. There were many questions on the Attivo acquisition, but ultimately management stated it would be accretive on gross margin and accretive on operating margin (whew – you could feel the collective sigh of relief on the call – we do not want those margins going in the wrong direction right now).

“So obviously, we have stated that [Attivo] accretive to our gross margin. From an operating margin standpoint, I think this current year, while we take into account our expected integration costs, the margin profile will be very similar to ours. If you take out those costs on a go-forward basis, it would be accretive to ours on an operating margin basis. And we'll have that guidance, obviously, when we specifically guide next quarter.”

Attivo’s specialization with Active Directory lends the acquisition to a post-SolarWinds world. Active Directory is the widely-used authentication system that was leveraged by hackers in the SolarWinds hack.

SolarWinds is known to be the “largest and most sophisticated operation that [Microsoft] has seen” with over 1,000 skilled engineers likely working on the operation. Although Crowdstrike was not infiltrated due to not using Microsoft’s email client, the cybersecurity company also did not detect the abnormal API calls until after FireEye disclosed their breach. The blog published by Crowdstrike said that they later discovered “abnormal calls to Microsoft cloud APIs during a 17-hour period several months ago.” In response, Crowdstrike released a reporting tool to manage permissions for Azure AD environments.

Crowdstrike has cited issues with the authentication architecture with Active Directory on Windows and Azure Active Directory, which was exploited to move laterally in the SolarWinds hack. The CEO of Crowdstrike called Microsoft’s Active Directory “antiquated.”

It may be a boon for SentinelOne that Crowdstrike is throwing shade on Microsoft as a large percentage of the Fortune 500 are committed to Microsoft. Notably, SentinelOne specifically stated in the past it covers email and will now cover Active Directory with best-of-breed identity threat detection and lateral movement prevention, which was used in SolarWinds.

The acquisition price is $617 million in cash and stock contributing $40 million in revenue for the full year. Attivo has ARR growth of 50% with primarily large enterprises making up the customer base with an analyst pointing out that ARR per customer for Attivo is double Sentinel.

Q4 Earnings Overview

SentinelOne leads their earnings calls with ARR growth as the top key metric for the company. In Q4, ARR growth was 123% and revenue growth was 120% year-over-year for total revenue of $65.6 million.

Total ARR is nearing $300 million while annual revenue for the upcoming fiscal year 2023 is guided at $368 million, with ARR suggesting this guide could be easily met over the next four quarters. Most importantly, customers over the $100K range are growing at a rate that is double overall customer growth at 137% and 70%, respectively.

The overall customer growth represents a slowdown from 79% YoY to 70% YoY while larger account growth was fairly flat at 141% in Q3 to 137% in Q4.

The company guided for Q1 revenue of $74.5 million, compared to revenue in Q4 of $65.6 million. This is important because management has stated in the past, Q1 revenue was down sequentially by 20% to 25%. “Our revenue guidance for Q1 implies that we should be at or better than typical Q1 net new ARR seasonality, which has been down between 25% to 35% sequentially in the past 2 years.”

Notably, the I/O Fund is unable to track where the ARR was down “for the past 2 years” but the sequential growth is headed in the right direction. The numbers we have show Q1 FY2021, net new ARR declined 37% QoQ to $8 million yet in Q1 FY2022 it grew +8% QoQ to $30 million. This year, the sequential growth will be +13.5%.

Higher ARR sequentially for the upcoming Q1 is likely driven by the record number of 100,000-plus deal and a record number of million-dollar plus deals. International is another area of strength as the company saw revenue grow 140%. This represents 31% of revenue – so something to watch closely as a near-term driver.

Net retention is higher at SentinelOne at 129% down from 130% compared to Crowdstrike in the 121-123% range.

In the current macro climate, across all growth stocks, the top line is battling the bottom line. On one hand, the market is trying to price in a slowing growth environment, and on the other hand, the market wants a perfect bottom line. Rarely do these two things coexist – strong growth in a slowing growth environment with a great bottom line. We are tech growth investors so we want to be careful of demanding that tech growth stops acting like tech growth.

What we are looking for at I/O Fund is what companies can maintain growth with an improving bottom line.an improving bottom line. We think to dump resilient growth in a slowing growth environment isn’t necessarily the correct answer. Obviously, we are well diversified with top holdings that have strong bottom lines (AMD, NVDA) but we are also not dismissive of growth-oriented business models.

With that said, SentinelOne has adjusted operating losses of ($63) million. The company has GAAP-operating losses of 109% or ($71) million. The margins noted include a 12% improvement to gross margin and a 38% improvement to operating margin.

It would be easy to discount the company based on the losses and to look at Crowdstrike with positive free cash flow and believe it’s the better stock. In this fierce debate, we are siding with SentinelOne primarily for its ability to run automated security from the data lake, as well as next-gen EDR/endpoints.

SentinelOne’s forward guidance is for 80% growth in FY2023 for $368 million in revenue and 99% growth for the first fiscal quarter ending in April for $74.5 million in revenue. The company is guiding for “high 60% gross margins by year-end” on a Non-GAAP adjusted basis and operating margins of negative (57.5%) operating margin by year-end for an improvement of 25%. The company’s long-term target is EBIT margin of 20%-plus.

Sales and marketing costs are improving over time with S&M trending closer to 100% of revenue in the past to 65% of revenue. Usually if an incumbent has strong margins like Crowdstrike, it serves as a model to help alleviate concerns of profitability long-term. Crowdstrike became profitable around the $200 million quarterly mark in July of 2020 so that could put us around H2 CY2023 for SentinelOne. Bradley goes into more depth below.

Notably, Crowdstrike had a rocky road with the stock being down 50% during the Q3 2019 selloff and then a total of 68% from peak to trough during March 2020. However, even if an investor had terrible timing and bought at IPO, the stock is now up 251% following yet another major selloff of Q1 2022. At its most recent peak, the returns were 390% if you had bought at IPO.

I’m sure you can see the parallel I’m drawing with SentinelOne and some of our more recent buys – which are at/near the low. In the fierce debate between Crowdstrike and SentinelOne, the only thing that matters to us is whether SentinelOne’s product can potentially carry the newly public company on a similar path in terms of price action. We believe it can.

Comments on SentinelOne’s margin relative to CrowdStrike’s

By Bradley Cipriano

SentinelOne and CrowdStrike are peers, so it makes sense to compare the two. However, CrowdStrike’s financials are much stronger than SentinelOne’s largely because CrowdStrike has significant scale (CrowdStrike is at nearly 6x the revenue run rate as SentinelOne). To account for the differences in scale, I compared SentinelOne’s latest results to CrowdStrike’s results when it was at a similar revenue run rate. As I’ll discuss below, SentinelOne’s results appear in-line with CrowdStrike’s but there are important differences in accounting treatment that impact the comparison.

SentinelOne’s Q4 FY2022 sales of $65 million are about equal to CrowdStrike’s Q3 FY2019 sales of $66 million. I use Q3 FY2019 for my base period for CrowdStrike, but also make comparisons to Q4 FY2019 to better account for seasonality.

At this stage, SentinelOne is similar to CrowdStrike based on both sales growth and gross margins. For instance, SentinelOne grew sales 17% QoQ in Q4 FY2022 vs CrowdStrike’s 19% QoQ growth in Q3 FY2019. SentinelOne’s gross margin was 63%, which was slightly below CrowdStrike’s gross margin of 66% in Q3 FY20219.

An area where SentinelOne is showing leverage is sales efficiency. SentinelOne’s Q4 FY2022 Sales and marketing expense was 64% of sales, which compares favorably to CrowdStrike’s 70% S&M margin (as of Q3 FY2019). However, Q4 tends to be a seasonally strong quarter for Enterprise, and relative to CrowdStrike’s Q4 FY2019 S&M margin of 61%, SentinelOne was slightly above that metric.

Further down the income statement, the differences grow between the two peers. SentinelOne’s R&D and G&A margins were much higher than CrowdStrike’s at this stage. For instance, SentinelOne’s R&D margin was 65% vs 39% for CrowdStrike, and its G&A margin was 42%, or double CrowdStrike’s 21% margin at a similar revenue run rate. As a result, SentinelOne’s operating margin of -108% as of Q4 FY2022 compares unfavorably to CrowdStrike’s -63% and -39% operating margin as of Q3 and Q4 FY20219, respectively.

However, there are important differences in accounting that impact this comparison. For instance, SentinelOne has accrued much more expenses than CrowdStrike at this stage, which essentially pulls forward expenses. For instance, SentinelOne’s accrued expenses increased 252% YoY to $84 million, which represented 61% of quarterly operating expenses. A rise in accrued expenses leads to a concurrent rise in operating expenses on the income statement, and is a sign of conservative accounting.

It appears that SentinelOne has more conservative accounting than CrowdStrike did at a similar run rate. For instance, in Q4 FY2019, CrowdStrike’s accrued expenses were 46% of total operating expenses. Had SentinelOne’s accrued expense profile been similar to CrowdStrike’s, it would have reported ~$20 million less in quarterly expenses. Stated differently, SentinelOne’s operating margin would have been closer to -78% in Q4 had it not ramped the accrual of expenses, which compares more favorably to CrowdStrike’s operating margin at a similar run rate.

Another important concept to consider is the capital intensity of both business models. If a company capitalizes more expenses to the balance sheet, its earnings will look relatively stronger. It is noteworthy that SentinelOne has just $25 million in net PP&E while at a $65 million quarterly revenue run rate, while CrowdStrike had $74 million in PP&E, or nearly 3x as much, at a similar run rate. During the latest quarter, SentinelOne capitalized just $1 million of expenses to the balance sheet, versus $15 million for CrowdStrike in Q4 FY2019. This $14 million delta impacts comparisons between the two.

Taken altogether, SentinelOne recognized ~$20 million more in accrued expenses, and CrowdStrike capitalized ~$14 million more in expenses while at similar revenue run rates. If we adjust SentinelOne’s earnings for this $35 million delta, its Q4 operating margin would be closer to -55%, which compares more favorably to CrowdStrike’s -39% operating margin as of Q4 FY2019 and -63% operating margin in Q3 FY2019.

To summarize, SentinelOne is at a similar run rate as CrowdStrike in Q3 FY2019. Sequential revenue growth is about even and gross margins are comparable. SentinelOne has moderately stronger sales leverage but has much higher R&D and G&A expense. However, SentinelOne has more conservative accounting, as it has expensed more costs to the income statement rather than to the balance sheet relative to CrowdStrike. For instance, had SentinelOne accrued expenses at a similar rate as CrowdStrike, its quarterly operating expenses would be ~$20 million lower. Furthermore, CrowdStrike has capitalized more expenses than SentinelOne has at a similar run rate, which added a further ~$14 million delta between the two. By accounting for these timing differences, SentinelOne’s operating margin appears more in-line with CrowdStrike’s at a similar run rate.

Finally, SentinelOne guided for 80% topline growth for FY2023, which is slightly above CrowdStrike’s guide for 74% topline growth for FY2020. SentinelOne is expected to grow slightly faster than CrowdStrike did at a similar run rate. The faster growth rate will front load more expenses, and also contributes to a lower margin profile. In all, SentinelOne’s margins appear relatively in-line with CrowdStrike’s after considering accounting differences and projected growth rates.

Marvell Technology, Inc. Update: Q4 FY2022

Posted on March 25, 2022June 30, 2026 by io-fund

Marvell reported strong Q4 results and is on the precipice of ramping demand from 5G and cloud infrastructure spending heading into FY2023. We have discussed throughout our coverage of Marvell and Inphi that the combined company is well positioned to benefit from these two trends.

Beth had outlined that “the growth opportunity for Marvell (and the reason I am investing) is for the lead Marvell currently has in 5G”, adding that datacenter will also be a strong tailwind for the company. With 5G ramping and datacenter growth expected to surge in the upcoming quarter, Marvell is well-positioned to benefit from these two secular trends. Importantly, Marvell has the inventory on hand to meet the rising demand, a trend that deserves a premium in today’s supply-constrained economy.

We expect that our thesis around 5G and datacenter growth will be realized in the first half of 2022, leading to outsized growth. Marvell’s financials have also been improving and the company will likely resume share repurchases later this year. While the outlook for Marvell is strong, we will be monitoring the back half of the year for a potential slowdown in growth as the 5G ramp peaks. However, this should be offset by strong demand from datacenters and edge computing tailwinds. I discuss Marvell’s latest results and drivers of demand in more detail below.

Marvell: Q4 Earnings Review and Outlook

In the latest quarter, sales increased 68% YoY to $1.3 billion, an acceleration from the 61% YoY growth rate in Q3 and beating estimates by 1%. During the year, Marvell merged with Inphi in a $10 billion transaction in Q1, and then also purchased Innovium for $1 billion in October. The recent M&A activity has skewed results, but on a Pro-forma basis, 2021 sales growth accelerated to 26% YoY, up from the prior year Pro-forma growth rate of 22%.

Sales were driven by strong growth in Datacenter and Carrier infrastructure (5G). Datacenter growth increased 113% YoY (15% QoQ) to $574 million or 43% of Q4 sales. On the Q4 call management explained that “the majority of the growth was from cloud, driven by robust demand from hyperscale customers.”

Carrier infrastructure increased 45% YoY to $241 million (18% of sales), and also accelerated on a sequential basis, growing 12% QoQ, up from 9% QoQ in the prior quarter. The strong growth was driven by Marvell’s 5G business, which grew sequentially by over 30% and exceeded management’s initial guide. On the call, management explained that it benefitted from the broader roll out of 5G technologies, and expects growth to continue into Q1 FY2023. I outline expectations for 5G spending in more detail below.

A rising trend for Marvell is its Enterprise Networking, which grew sales 64% YoY to $263 million. Management explained on the Q4 call that this end market was “going through an inflection” as hybrid work environments are driving demand for an “extended period of refreshing [enterprise] infrastructure”. This includes increasing bandwidth and improving security. To be complete, automotive increased 134% YoY to $79 million and Consumer increased 11% YoY to $185 million.

Gross margin declined YoY from 53% to 50%, which was driven by a step-up valuation in inventory and other non-cash charges following the M&A activity during the year. Non-GAAP gross margin increased 160 bps YoY to 65%, which is high relative to peers. Following the rise in non-cash expenses, operating margin fell YoY from 5% to 3% while non-GAAP operating margin expanded 860 bps YoY to 36%. GAAP earnings were $0.01/share, which missed estimates by a penny while non-GAAP EPS of $0.50 beat estimates by $0.02 and grew 72% YoY.

Cashflows from operations were $346 million and net leverage was reduced to 2.3x. Management explained on the call that they are on track to reach their targeted leverage ratio of 2x by the end of Q2, at which time they expect to restart share repurchases. This is a favorable trend and could support a higher share price, all else equal. Prior to the Inphi merger, nearly 100% of free cash flow was directed towards share buybacks. Utilizing Management’s long-term guide for 32% FCF margins, Marvell has capacity to lower its share count by about 4% per year, which will accelerate EPS growth.

Looking forward, sales are expected to accelerate and grow 71% YoY (6% QoQ) to $1.425 billion. Datacenter sales are expected to grow over 100% YoY in Q1 and in the mid single digits on a sequential basis, while Carrier infrastructure sales are expected to grow over 40% YoY and in the low-single digits on a QoQ basis. Enterprise networking sales are expected to accelerate and grow in the mid-teens on a QoQ basis while Automotive is also expected to remain strong and grow QoQ in the high-single digits. Finally, Consumer is expected to be flat QoQ. Non-GAAP EPS growth is expected to accelerate and rise 76% YoY to $0.51. I discuss the core drivers of Marvell’s demand in more detail next.

Update on 5G infrastructure

As outlined in our initial analysis, Marvell has a leading position as a 5G supplier and supplies the components for 5G base stations to customers such as Nokia and Samsung. We expected that Marvell would take a commanding lead in 5G infrastructure in 2021, a trend that we can see has finally arrived, as Marvell’s 5G sales increased 30% QoQ in the latest quarter.

Marvell’s 5G customers include Nokia and Samsung, which partner with carriers such as T-mobile, Verizon and AT&T to build out their 5G infrastructure. As a result, looking at capex plans from these telecoms can provide insight into the expected ramp in 5G spending going forward. As I’ll highlight below, the big three American carriers expect to ramp spending on 5G deployments by about $10 billion in 2022, which will drive demand for Marvell’s products.

Verizon explained on their Q4 call that they expect incremental capex related to the 5G upgrade cycle to peak this year and then start to normalize. Specifically, Verizon CFO Matt Ellis explained that Verizon had guided for “incremental [5G CapEx of] $10 billion over five years. We're going to see the biggest part of that come through this year”. In the chart below, Verizon highlights that its C-Band overlay spending will ramp in 2022. C-band is the wavelength that Verizon is using in its 5G deployments. Importantly, the accelerated $10 billion in 5G spending is expected to conclude in 2023, so the 5G ramp will be a quick, but significant trend for Marvell.

Verizon Outlook for Capital Expenditures

Source

AT&T also guided that its capital expenditures are expected to ramp in 2022 and 2023. AT&T’s CEO stated during the company’s 2022 investor (03/11/2022) day that “it's a race to the home and to deploy 5G across the country. Our capital investment will be elevated over the next few years as we aggressively build a next-generation network with fiber and 5G”. AT&T’s guide for CapEx is expected to rise to $5 billion in both 2022 and 2023, driven by 5G deployments (shown below).

AT&T Outlook for Capital Expenditures Capex guide

Source

Finally, T-mobile recently increased its capex guide for 2022 to maintain the company’s position in 5G. T-Mobile’s capex has grown from $6 billion in 2019 to over $12 billion in 2021. Looking forward, T-mobile expects its capex to continue to rise to another $2 billion and reach around $14 billion in 2022 as it pulls forward 5G spending.

In aggregate, Verizon, AT&T and T-mobile are guiding for a ~$10 billion (~16%) rise in capital expenditures in 2022, mostly driven by the deployment of 5G infrastructure. Nokia, which is a significant customer of Marvell’s, explained during its Q4 earnings call that spending plans from American telecoms is a favorable trend. Specifically, management stated that “listening to the CapEx plans of the key [telecom] customers in America that is of course a reason for optimization”.

We expect that 2022 will be a peak year for 5G spending, directly benefitting Marvell. However, trends in datacenter are long-term secular trends that should sustain topline growth beyond 2022. I discuss this in more detail next.

Update on datacenter

With 5G ramping and likely peaking this year, Marvell will also be benefitting from another trend that is just now beginning to ramp: COLORZ II. We had outlined COLORZ in our Inphi analysis, explaining that “Inphi’s COLORZ silicon photonics technology allows data centers in the same metropolitan region to function like a mega data center. This facilitates faster edge computing within an 80/120 km distance for 30-megawatt data centers as they will be linked together and function like a 120-megawatt data center … When the COLORZ ZR 400G launches, it has the ability to become a critical supplier for data center interconnects and the converged edge of telecom and cloud connections.”

The time has arrived for the ramp in COLORZ 400G ZR. Management explained on the Q4 call that it expects datacenter revenue (its largest segment) to increase more than 100% YoY driven by the “strong ramp” in its 400-gig ZR datacenter interconnect products, which is termed COLORZ II.

CEO Matt Murphy added that the first iteration of COLORZ peaked at a $100 million run rate, which was driven primarily by one customer (Microsoft). In the upcoming iteration of COLORZ II, revenues will surpass the prior peak of $100 million in Q1 and will continue to grow from there. This is because the 400ZR is being adopted by multiple hyperscale customers, so revenues will be more substantial.

To get a sense of the cadence of topline growth we can expect from COLORZ II, we can use Inphi’s prior financials to provide a guide. COLORZ first shipped in volume in 2017 and Inphi recorded $59 million in sales from COLORZ in 2017, which then grew into $89 million in sales by 2020 and eventually peaked at a $100 million run rate, or nearly doubling overtime. CEO Murphy is saying that COLORZ II will start at the $100 million run rate and continue to grow thereafter. Considering COLORZ II has multiple customers, the ramp should be even more robust than the first iteration and COLORZ II sales could more than double overtime.

Since Marvell is directly tied to datacenter infrastructure spending, a decent proxy for demand is trends in CapEx from leading cloud providers such amazon AWS, Google Cloud and Microsoft Azure. As shown below, AWS, Azure and Google Cloud have ramped spending, and this spending is expected to continue to grow.

Specifically, Amazon stated during its Q4 call that “Just under 40% of that CapEx is going into infrastructure, most of it’s feeding AWS … If I look to the future, we’re still working through some of our plans 2022, but it’s coming into focus a bit. We see the CapEx for infrastructure [AWS] going up. We still have a very fast-growing business thats growing globally, and we’re adding regions and capacity to handle usage that still exceeds revenue growth in that business”

Google stated during its Q4 call that “In 2022, we expect a meaningful increase in CapEx.” And Microsoft added that it expects capex will be up YoY in the upcoming quarter. The increased capital expenditures from cloud providers is a favorable trend that will benefit Marvell’s topline going forward.

Marvell has the inventory to meet demand but there are risks

Given the expected ramp in Datacenter and Carrier infrastructure sales, which collectively accounted for over 60% of Q4 sales, it is important that Marvell has the necessary inventory to supply this demand. Marvell’s inventory levels have increased sharply recently, and rose 169% YoY to $720 million, outpacing the 68% YoY rise in quarterly sales. Moreover, since Marvell has the inventory on hand, it backs up management’s statements that they expect a significant ramp in revenue in the near term.

However, having excess inventory is typically an unfavorable trend, since the technology can quickly become obsolete which leads to lower prices, impacting earnings. This concern is somewhat offset by the scenario outlined above about ramping demand from both 5G and datacenters, suggesting that the build-up of inventory is appropriate. Furthermore, Marvell’s inventory composition is relatively healthy and is not loaded with idle finished goods inventory, which is at a higher risk of being written off. As shown below, finished goods were just 20% of total inventory, well below the five-year average of 31%.

A key risk that should be noted is the way Marvell sells its inventory. Marvell does not have agreements in place that guarantee sale to its customers. Marvell explained in its 10K that it must maintain large inventory balances because the “semiconductor industry is characterized by short lead time orders and quick delivery schedules”. If demand for its products declines, Marvell will be left with a very large inventory balance that will likely need to be discounted to turnover.

Another risk is that Marvell has now had to enter into manufacturing supply capacity reservation agreements with foundries to secure supply. This means that Marvell now has to prepay for inventory (unfavorable) and also must pay a fee to cancel its reserved capacity. Marvell has $2 billion of supply commitments signed through 2032 and prepaying for future supply increases Marvell’s risk of taking on too much inventory, which could pressure margins in the future. This is a relatively new development and is a direct result of the current supply chain shortage. However, this is broader trend in the semiconductor industry and is not isolated to just Marvell. Furthermore, securing supply in a tight market should be awarded a premium.

The key takeaway is that elevated inventory can be a concerning trend, but we think it is actually a favorable trend given the expectations for surging demand in the near term and supply-constrained environment. Furthermore, inventory composition is healthy with low levels of idle finished goods. We should expect to see Marvell’s inventory balance normalize going forward as the 5G and COLOR II ramp get underway.

Trends on the horizon: DPUs and customization

With datacenter and 5G taking center stage in 2022, there is a new trend gaining momentum that promises to be a significant driver of growth going forward that should offset the eventual decline in 5G spending. The rising trend is ‘customization’, which is being driven by hyperscale customers that are increasingly developing their own custom, optimized silicon for the cloud environment.

Marvell’s recent acquisition of Innovium was driven to improve the company’s reach in the cloud-optimized market. Innovium developed a leading cloud-optimized switching technology that is used in cloud data centers. Innovium is expected to report $150 million in sales in FY2023 after being selected as a supplier for a Tier 1 cloud customer. Marvell also disclosed that it has a strong pipeline of cloud-optimized silicon, with $400 million of contract wins in the pipeline that will turn into revenue in FY2024 which is expected to double to $800 million by FY2025.

Since the cloud environment is inherently different than the legacy on-prem environment, prior architectures developed long before the cloud was around are outdated and are not optimized for the cloud. It makes sense that new silicon solutions will be optimized specifically for the cloud, and Marvell is positioning itself to benefit from this rising trend.

For instance, Marvell disclosed in its 10K that it is transitioning its product offering “from standard server processors to the broad server market to focus only on customized server processors for a few targeted customers”, adding that “the demand for optimized solutions has been increasing as our customers seek greater customization and differentiation for their products and services”.

Customized silicon for the cloud will be a material contributor to Marvell’s topline in a few years, and will help offset the expected decline in 5G spending after the ramp in 2022. During the year, Marvell won “over a dozen cloud optimized programs across multiple Tier 1 cloud customers. A significant number of these designs are for custom DPU implementations, reflecting the increase in the attach rate of DPUs inside cloud data centers”. We had mentioned that DPUs would be a tailwind for Marvell, stating that Marvell will be a major player here and this trend will be a future bull thesis for Marvell.

Marvell is well-positioned to benefit from the rise in edge computing driven by 5G and datacenter growth, and new trends such as custom, cloud-optimized silicon. We expect to hold onto Marvell through the concurrent ramp in 5G and datacenter spending but will monitor the company closely heading into H2 2022 for a possible lull in growth as spending peaks. however, we expect datacenter, edge computing and AI tailwinds to drive topline growth for the foreseeable future.

Valuation and conclusion

Marvell trades at a 10x forward P/S multiple, down from its peak multiple of 17x in December 2021, but a premium to peers such as Broadcom (8x), AMD (7x), Qualcomm (4x) and Cisco (4x). Marvell also trades at a 30x forward PE multiple, down from its peak of 57x in December 2021 but also above the peer median of 17x. Importantly, Marvell’s Q4 earnings grew 72% YoY and are expected to accelerate to 76% YoY growth in the upcoming quarter. The company’s long-term guide for 40% operating margins going forward, coupled with its LT guide for ~25% topline growth, highlights that earnings growth will be significant going forward, a trend that supports a premium PE multiple. While Marvell trades at a premium relative to semiconductor peers, the company is well-positioned to benefit from strong secular tailwinds such as 5G, datacenter, AI and customized silicon, which warrants a premium multiple.

We believe that the upcoming year will be a breakout year for Marvell’s top and bottom-line as both 5G and COLORZ II are ramping this year. Marvell also has the inventory on hand to meet this demand and is positioning itself to benefit from new trends such as customized silicon. While we believe that Marvell will be strong, we will be monitoring growth expectations closely heading into H2 2022 as sales growth may slow as 5G spending peaks. however, we expect this to be a temporary trend that will be offset from the secular tailwinds from datacenter and edge computing infrastructure spending.

Additional Reading:

Disclosure: Bradley Cipriano and the I/O Fund own shares in Marvell and have no plans to change their respective positions within the next 72 hours. You can access the I/O Fund’s positions here. The above article expresses the opinions of the author, and the author did not receive compensation from any of the discussed companies.Disclosure: Bradley Cipriano and the I/O Fund own shares in Marvell and have no plans to change their respective positions within the next 72 hours. You can access the I/O Fund’s positions herehere. The above article expresses the opinions of the author, and the author did not receive compensation from any of the discussed companies.

Confluent Update and Q4 Earnings

Posted on February 24, 2022June 30, 2026 by io-fund

Below, we do another overview of Confluent’s product and an update following the Q4 earnings report. Here are two resources we recommend reading from our premium site for more information on the company.

Confluent Product Overview and Q3 Earnings

Big Data, Analytics and the Importance of ML

We believe open source with enterprise-grade features will become a key market moving forward as it solves for the downside of open source such as a lack of technical support. In Kafka’s case, the downside are things like a lack of data verification and having to manually connect to various data warehouses and other platforms to import/export data. Confluent also makes the argument that multi-cloud and hybrid cloud architectures are best served with a supported enterprise version for multi-tenancy security and data residency.

Notably, from my perspective, we are not betting on Confluent being used over the open-source version of Kafka in a direct competition, rather we are betting that Kafka will increase in importance. In this case, if Kafka continues to grow, Confluent will take a percentage of this market share should more enterprises prefer a managed version of Kafka. 70% of the Fortune 500 use Kafka and 80% of the Fortune 100. According to this site it has a 12.5% market share.

Kafka is popular because of its high-performance real-time data streaming capabilities for mission critical applications. It is distributed and fault-tolerant, which means if one component fails, the system will still work. It can also scale to hundreds of clusters and billions of messages.

As discussed in our original write-up, Kafka was developed at LinkedIN to process the large number of messages per second the social media company handles. The framework enables event streaming, which helps messaging and data integration. There is high scalability with a publish/subscribe model that allows applications to share and create data in a serverless and microservices architecture. What Kafka solves for is the ingestion of events data in real-time with low latency with continuous read/write. If data remains at rest and/or in a mainframe environment, then companies cannot be truly data-driven. Kafka on the other hand can scale from a billion messages per day to a trillion messages per day.

Machine Learning and Kafka

Confluent opens up the amount of data that can integrated. The thesis is the increase in the number of companies that will need real-time data processing and real-time data analytics due to the increase in software driven architectures. The idea is that “data in motion” will replace data at rest, or batch data processing from traditional databases. This is also important for the real-time data streams that machine learning requires.

Kafka is more than a messaging system as discussed in this article and is used for business applications, streaming ETL middleware, real-time analytics and edge/hybrid use cases for the framework.

Here are some examples of how Kafka can be used outside of messaging systems:

Fraud detection through a machine learning pipeline for Paypal’s billions of messages
Data correlation in real-time for Lyft for matching maps, estimated time of arrival and cost calculations
Unity uses Confluent to be internally data-driven across R&D and cloud-services, plus to help drive the monetization network by rewarding players for watching ads and incorporating banner ads
Continuous calculations for betting platforms
Drug discovery that is automated and scalable

Machine learning requires model training from historic data and also model deployment for scoring and predictions. Training can be done with batch yet scoring is partial towards real-time data. ML-powered applications run inferences on large volumes of data to return predictions very quickly (milliseconds). Rather than use Remote Procedure Calls (RPC) and frameworks like gRPC, some companies use a Kafka streaming model.

Here is how the company states the problem that Confluent seeks to solve:

“By becoming more software driven, more businesses will rely on real-time data. Confluent believes that data in rest is not able to meet the current and future demands of software-driven businesses. Daily batch processing and static real-time queries or “point-in-time” queries with stored data lead to an unnecessarily large and tangled architecture that is not capable of data flow between applications.”

Enterprise-grade Features

As with Spark and other open-source projects, there is a marketplace for making the frameworks easier to use. Confluent Kafka opens up the amount of data that can be integrated, for example, to combine transactional data (orders, inventory) with sentiment-driven data (likes, page clicks). This helps with predictive analytics and also machine learning because the “data flow” allows for algorithms to work as they are intended to.

In order for data to be in motion, Confluent’s platform connects data from many different sources. The company has over 50 fully managed connecters with Big Data and Analytics from Azure, Amazon/AWS, Google and Databricks. Without these connectors offered by Confluent, integrations between systems on an open-source framework can take months and also require intensive resources to manage.

Confluent is attempting to stave off competitors through “completeness of product” which touches on our multi-cloud and hybrid cloud discussion. We’ve discussed hybrid for a few years, yet our most recent write-up was here and here on Datadog. The recent write-up is worth a read if you want to know exactly why agnostic, best-of-breed products are sometimes outpacing Big Tech when it comes to cloud services and products. Datadog is the best example of a product where customers are avoiding vendor lock-in.

The completeness of product goes beyond multi-cloud and hybrid as Confluent is attempting to hold off competitors through data security and data governance, as well. Because data is often an organization’s most prized asset, it often has internal processes for compliance. There is often external, geographic compliance required by governments and industry agencies, as well, for global companies.

In order for completeness of product to work, Confluent needs to have a large geographic footprint. The company has added eight more regions for Confluent Cloud with an emphasis on APAC. There is also a new partnership with Alibaba Cloud. This can help offer differentiation for multinationals who have operations in China.

Competitors:

Regarding direct competitors, one example is Amazon MSK which offers a competing managed streaming service. This competitor is a good option for developers provisioning a Kafka cluster and a new streaming platform may not be needed in this case.

Rather than re-architect Kafka to be cloud-native, Amazon MSK cloud-enabled it as provisioned infrastructure. This means Confluent is stronger than MSK with scaling elastically by offering elastic quotas, which eliminates the need to size clusters for spikes. It’s also stronger on multi-tenancy security. Amazon MSK also does not offer Kafka Connect or Kafka Streams.

For more enterprise uses where Kafka Connect or Kafka Streams is required, then Confluent is more likely to be used to save development time and learning curve in writing Kafka Connects sinks and source.

Blockchain and Metaverse Potential

We’ve written at length about Confluent’s core use. However, there is a blockchain potential with Confluent with one case study right now with Dapper Labs.

“These are steps that attracted Dapper Labs. They're one of the most innovative NFT companies delivering fun and games on the blockchain. They have a number of decentralized apps, but one that's risen dramatically in popularity is called NBA Top Shot. To date, there have been over 10 million digital collectible transactions and Confluent is at the center of their data streaming architecture to facilitate these purchases. Dapper chose us to run their mission critical workloads because of the scalability and security of our cloud solution.”

There’s also a case for 5G networks needing data in motion. Here’s what was said about Dish on the call:

“A significant customer for both AWS and us is DISH Network. With their new 5G smart network, DISH is transforming how people and enterprises leverage data. They deployed Confluent Cloud over AWS to connect their network systems and customers with real-time data. This means that Confluent is a key part of their network's data backbone, starting with fault management and network resiliency functions to ensure network availability, and our enhanced collaboration with AWS is making it easier for customers like DISH to unlock data in motion everywhere.”

Confluent Q4 Overview

Confluent has been accelerating in revenue for four consecutive quarters and also across other key metrics.

The company reported fiscal year 2020 revenue growth of 58% year-over-year and fiscal year 2021 revenue growth of 64% year-over-year. Confluent Cloud revenue growth for fiscal year 2020 was 117% compared to FY2021 revenue growth of 200% year-over-year.

If we look at Q4, total revenue is outpacing the fiscal year growth for 2021 and also outpaced Q3. Revenue growth for Q4 was at 71% — the highest growth rate from publicly available information which dates back two years to Q1 2020.

Cloud revenue did decelerate on a sequential basis, however, the company stated Q4 is often seasonal due to engineers being out of the office and on vacations. We will see if this picks back up in Q1. Regardless, on an annual basis there was a significant improvement. Notably, if we look at 2020 cloud revenue, we can see it’s lumpy at times with Q3 2020 being the weakest and Q2 2020 being the strongest.

In regards to “sandbagging” which is essentially the company guiding low and blowing out the guidance, which has happened a few times now, the company has a lot of moving pieces in terms of business model and likely wants to win trust with institutions. We are not opposed to this even if it means the price action was somewhat severe after the earnings report due to the guidance. What we are more concerned with is that Confluent continues to raise and beat, and that the underlying key metrics help us to substantiate the company’s longer-term strength.

Bradley stated the following in our last write-up and got pretty close to the revenue growth that Confluent actually reported:

Looking forward, management guided that Q4 revenue will rise 55% YoY $109 million, which would mark a deacceleration from the most recent growth rate of 67% YoY growth. However, this estimate is likely conservative, as management guided that Q3 sales would grow 46% YoY to $90 million and actual Q3 sales grew 67% YoY to $103 million. If we assume that Confluent beats it guide by a similar amount in Q4 as it did in Q3 ($13 million), then Q4 sales growth will accelerate to 73% YoY (this is merely an observation – no guarantees).If we assume that Confluent beats it guide by a similar amount in Q4 as it did in Q3 ($13 million), then Q4 sales growth will accelerate to 73% YoY (this is merely an observation – no guarantees).

Most notably, the company is reporting high remaining performance obligations growth of 91% year-over-year. This is higher than the 75% year-over-year we saw in Q3.

Bradley discussed this in our last write-up:

Confluent also states that RPO is an important metric to monitor in order to measure the health of the sales pipeline. In Confluent’s first conference call as a public company (Q2), CFO Steffan Tomlinson explained that:

“Given the various revenue components and billing terms in our model, remaining performance obligations or RPO and current RPO rather than billings, are important metrics to measure the health of the business. RPO provides insight into the organic momentum of our business as it represents contractually committed revenue to be recognized in the future regardless of billing terms and variability in cloud consumption pattern”. RPO provides insight into the organic momentum of our business as it represents contractually committed revenue to be recognized in the future regardless of billing terms and variability in cloud consumption pattern”

Financials Deep Dive

By Bradley Cipriano

A slight blemish during the quarter was Confluent’s customer growth, which lagged the growth in sales. Customers increased 65% YoY to 3,470, which lagged the 71% YoY growth in total sales. This drove subscription revenue per customer up 4% YoY to $31,000/customer, implying the recent acceleration in sales was driven by higher spending rather than customer growth.

Generally, growth from new customers is more sustainable and higher quality relative to growth from increased spending. However, DBNRR remained robust at over 130%, signaling that customers are increasing their spend over time.

It is odd that customer spending increased but cloud growth deaccelerated during the quarter. Since cloud is a usage-based revenue model, increased spending should have driven cloud outperformance. However, cloud spending slowed from 245% YoY growth in Q3 to 211% in Q4. On the Q4 call, management explained that cloud was impacted by seasonality due to relatively lower spending over the holidays which lead to slightly slower rates of usage. While this may be true, it doesn’t explain the YoY deacceleration, as this trend would have existed in the year-ago quarter. Nevertheless, there is inherent variability in a usage-based model so investors should not expect an acceleration in sales every quarter.

Given the slowdown in customer growth and slight deceleration in cloud sales, the Street may be concerned that Confluent’s growth may be somewhat cannibalistic. This would explain the sell-off in its stock following otherwise strong results which reported a beat and raise. Investors may be wondering if cloud growth is coming at the expense of platform growth, or vice versa?

CEO-Founder Jay Kreps discussed this concern on the call and stated that the company is growing both in the cloud and in hybrid environments. He said that “we don't really view this as kind of a transition where we're just shifting from platform to cloud and just kind of swapping out customers from one product to the other. Effectively, we have to have kind of an outpost in each environment a customer is in. So, we expect to continue to see growth in Confluent Platform throughout this, and we think that's not a bad thing. That's a good thing.” CFO Steffan Tomlinson added that “what our customers are telling us is, by and large, they're running hybrid environments”.

A common issue with ramping cloud sales is that sales in other parts of the business stagnant, but we do not believe this is the case. For example, Confluent’s financial results remain high quality which suggests that cloud/platform sales are not cannibalistic.

For example, net deferred revenue (deferred revenue less accounts receivables) increased 105% YoY to $109 million, or 31% of TTM subscription sales. This was an improvement from the 26% and 23% level in Q4 2020 and Q4 2019, respectively. The rise in net deferred revenue relative to subscription sales signals that the company is receiving relatively more cash upfront, improving the quality of topline growth. If sales were cannibalistic, we would have likely seen a reduction in cash receipts and/or a deacceleration in growth. Instead, cash improved and sales accelerated.

Furthermore, RPO also increased 91% YoY to $501 million, an acceleration from the 75% and 72% YoY growth rates in Q3 and Q2, respectively. While we need the 10K to fully assess the quality of RPO, total RPO represents 92% of management’s NTM guide, up from 81% in Q3. This improves the quality of forward sales and suggests that there is conservatism in management’s forward guide.

However, we do note that cash support for RPO declined slightly during the quarter. Total deferred revenue-to-RPO fell from 52% in Q3 to 49% in Q4. This trend is likely driven by the rise of cloud bookings, since cloud is a usage-based model and new cloud customers are typically on pay-as-you-go plans, which are billed in arrears. On the Q4 call, CEO-Founder Jay Kreps explained that cloud accounted for 50% of ACV bookings in Q4, highlighting how cloud will be the majority of revenues going forward. As customers become more familiar with Confluent’s products, they will likely increase their commitments and convert from pay-as-you-go customers to larger customers that pay upfront. As a result, we view the slight decline in upfront cash receipts as a natural progression for the firm and not a major concern at this time.

Cash Levels and Stock Based Compensation

Confluent recently raised nearly $1 billion in cash following a convertible debt offering in December. Following this raise, the company has over $2 billion in cash, which is well above its current cash burn of ~$108 million (based on TTM free cash flow). The company is focused on growth, so investors should be prepared for continued losses and cash outflows. On the Q4 call, management highlighted that their near-term priorities are to continue to invest in innovation and to expand its geographic footprint, signaling that growth is being prioritized over near-term profitability.

Nevertheless, given Confluent’s relatively large cash balance, we likely should not expect an equity raise in the near term. However, the company will still be dependent on capital markets until it is sustainably cash flow positive. Looking forward, the Street expects EBITDA (a proxy for cash flows) to remain negative through at least FY2023, suggesting that Confluent will remain reliant on capital markets for the next few years. Importantly, there are signs of improvement, as free cash flow margin improved from -30% in the prior year to -22% in the current quarter.

Furthermore, Confluent has relatively high levels of stock-based compensation (SBC), which subsidizes cash used for working capital but dilutes shareholders. Stock-based compensation has trended near 48% of quarterly sales for the last two quarters and was 40% of TTM sales. This is relatively high and ranks in the top 10 for cloud (shown below), but is a function of Confluent recently going public (which frontloads SBC). We expect SBC to decline as a percentage of sale going forward as it laps the IPO and topline growth outpaces expenses.

Nvidia Stock: How to Value the Metaverse

Posted on February 21, 2022June 30, 2026 by io-fund

Nvidia On How The Metaverse Can Overtake The Current Economy

This article was originally published on Forbes on Feb 18, 2022, 12:57pm ESTForbes on Feb 18, 2022, 12:57pm EST

I had the opportunity to talk with Richard Kerris, vice president of the Omniverse development platform at NVIDIA. In the interview, I asked Kerris questions that are on every investor’s mind, including pointed questions about where realreal revenue growth will come from, how large is the addressable market exactly, plus what CEO Jensen Huang meant when said that the “Omniverse or the Metaverse is going to be a new economy that is larger than our current economy.”

The Metaverse is particularly challenging for investors as the opportunity is enormous yet getting the timing right and also choosing which companies will participate in the futuristic yet burgeoning virtual economy will not be easy.

Here are the main points we discuss in the video:

Why the Metaverse Can Exceed Our Current Economy
How Universal Scene Description provides the portal to the Metaverse
The Importance of Ray-Tracing and the RTX Platform
Industrial Virtual Worlds

Watch the Nvidia Omniverse interview with Beth Kindig and Richard Kerris:

Candid Interview with Nvidia: Can the Metaverse Drive Real Revenue Growth

Last year, Nvidia CEO Jensen Huang said, “Omniverse or the Metaverse is going to be a new economy that is larger than our current economy.” Many investors look for large addressable markets, so to hear a CEO state that a technology could surpass the size of our current economy is a statement to pay close attention to. In the interview, I asked Kerris what Huang meant, and Kerris’ answer was quite simple: the Metaverse will exceed our current economy because it will be many times larger than the internet.

Here is a direct quote:

“[The Metaverse] is going to be many times bigger than the web because of what a virtual world can do for business, for education, for medical, for all sorts of things including entertainment; we’ve just begun to scratch the surface of these possibilities […] You’ve probably heard the term digital twin. One example is what it’s going to do to revolutionize the industrial market, design and manufacturing. Well, a digital twin is a true-to-reality twin in synthetic worlds of what happens in the physical world. We are seeing this transform these things because when you can make decisions in that synthetic world before you commit to it in the physical world, you have a lot of cost savings.” –Richard Kerris

According to Kerris, in simple terms, the Metaverse is virtual worlds. Rather than being a Ready Player One type experience where you are strapped down with heavy equipment, the Metaverse will instead offer more 3D experiences or environments that replace our current digital experiences.

The Real Value of 3D Virtual Worlds

Industrial 3D environments are especially ramping up as 3D virtual worlds result in cost savings, are safer for employees and allow for more iteration on designs. For example, BMW manufactures 2.5 million cars per year, and the company has made a digital twin of its factory to reduce any downtime when the company has to change its process for a new model. In the physical world, a new model affected production whereas now the company can change its process in a synthetic world to eliminate errors and downtime. The new approach helps BWM view their entire factory in simulation mode with photorealistic detail. Another advantage is that data is available immediately and any changes can be made in the planning stage itself, which saves time and money.

Another example as to the value of virtual worlds is route planning. Factories can perform route planning in a virtual warehouse in Nvidia’s Omniverse that allows orders to be optimized for automated order picking scenarios. Essentially, warehouses are able to reoptimize factory floor operations as new orders come in or as robots go offline. This optimization training can also save quite a bit of money.

Universal Scene Description (USD) Provides the Portal to the Metaverse

The portal to the Metaverse is Pixar’s Universal Scene Description. USD is hailed as one of the most important tools for building the Metaverse as it allows for a persistent experience to where the Metaverse is more connective and has consistency. Metaverse experts tend to reference the internet as a baseline for how the Metaverse will become widely adopted. In this Web 3.0 reference, Pixar’s USD functions like the connective HTML of 3D worlds.

“In the early days [of the web], once we started connecting things, we didn’t realize just how big it could be … one of the things that happened at that time was HTML. HTML allowed for a consistent plumbing or consistent connective tissue between the websites so you didn’t have to have a specific browser or a specific extension installed. Now, we go from one website to another and the videos play the same, the text is great, the images, etcetera. Much of the same kind of thing is going to happen with these virtual worlds because once you can teleport from one virtual world to another virtual world, and the experience of the worlds is about the content instead of the lighting and the materials, we will see a tremendous opportunity take place.” –Richard Kerris, on Universal Scene Description (USD)

Sign up for I/O Fund's free newsletter with gains of up to 1100% – Click here
USD)

Sign up for I/O Fund's free newsletter with gains of up to 1100% – Click here
Click here

With persistent experience, the Metaverse can rival Web 1.0 and Web 2.0, as it allows for virtual worlds to break out of siloed experiences. The USD open-source framework allows for the exchange of 3D data and has become a universally supported plugin for developers and creators to use as a baseline.

Nvidia’s Omniverse supports the USD plugin as a portal to access Nvidia’s rendering and graphics products. Apple, Autodesk, and many other companies are also using the framework to create a persistent experience for developers and creators. For example, the iPhone’s LiDAR scanner allows you to scan objects in USD. The scanned images are saved in a USDZ compressed file to be used across other platforms.

How Real-Time Ray Tracing Catalyzed Virtual Worlds

Nvidia was originally a hardware company that transitioned to also become a software company, which uniquely positioned the company to deliver ray-tracing to the Metaverse market. Ray-tracing allows for the simulation of light and physics to render graphics, resulting in a much more realistic and complex virtual world. After commercially releasing real-time ray tracing on a chip, Nvidia was then able to leverage RTX GPUs and ray tracing extensions for the renderer in their software simulation platform Omniverse.

“There was the big boom moment that took place for all of these things to happen. The first part of that was RTX, which is real-time ray tracing on a chip. It was really a milestone that will be forever remembered. Ray-tracing allows for photorealistic rendering to happen in real time. That means that the worlds we are talking about look more and more indistinguishable from the real world […] It was such a moment at Nvidia when that happened. I remember the enthusiasm of the computer graphics community. For those of us who have been in [the industry] for awhile, we dreamed of this 30 years ago. We knew it was that impactful of a moment.” -Richard Kerris

Ray-tracing renders 3D models through the physics of light. Through the physical simulation of light, subsurface scattering and light diffusion helps computer vision see things similar to how a human retina works.

Companies like Pixar have been able to do 3D rendering and computer-generated imaging for some time in films, television and gaming, however, Nvidia improves this process by adding the real-time capabilities to ray tracing through the RTX platform. Formally, it could take weeks to render animation that is only a few seconds long whereas the RTX platform reduces this time by offering more real-life retina vision. The RTX platform also offers better path tracing and denoising, which was first used for gaming, and now lends itself well to the virtual worlds of the metaverse.

More on the RTX Platform

The RTX platform provides APIs and SDKs running on Turing GPU architecture for applications to be built with ray tracing and AI-enhanced graphics. More than 200 games and applications use RTX including Minecraft, Fortnite and Cyberpunk. According to Nvidia’s previous earnings call, an estimated 25% of the installed base had adopted RTX GPUs. This is due to an overhaul on GPU gaming architecture as Nvidia has combined Turing RTX GPUs with Ampere RTX GPUs. This allows for AI rendering called Deep Learning Super Sampling (DLSS) to be combined with ray tracing. The success in gaming will only help the proficiency of real-time ray tracing and AI enabled super resolution capabilities for the Metaverse and 3D virtual worlds.

Industrial Virtual Worlds

Autonomous driving is one of the more compelling use cases for the Metaverse given the number of times autonomous vehicle systems were deployed on the roads and yet were pulled from the market due to accidents. Meanwhile, attempting to train models in the physical world (or analog world as Kerris puts it) can take a very long time.

“If we go back to the concept of a digital twin, a digital twin doesn’t have to be a factory, it can be a city. We have companies like Ericson who have built digital twins of cities for antenna propagation for the 5G network […] we’ve been using digital twins in that environment to have a synthetic world that these cars can be trained on that is indistinguishable from the real world, it’s exactly the way things are laid out, but you can have many cars being trained in that world at the same time. What’s most important is you can throw all kinds of predicaments at it so that it learns [..] that those cars in the analog world may never have encountered [..] there’s a much higher degree of confidence in how a [synthetically trained vehicle] will respond.”

The Isaac Sim toolkit is a robotics simulation and synthetic data generation tool that helps increase accuracy for robots. It supports a SDK and robotics operating system frameworks package to develop robotics AI and navigation applications. This helps to improve AI-based computer vision by improving the data sets, which in turn are used to train robots for increased understanding of their surroundings. The end result is fewer accidents and less human intervention.

Conclusion

It’s understandable if investors are skeptical of the Metaverse as the technology is essentially in the early adopter stage. My conversation with Nvidia helps to break down these walls around the virtual economy and how big it can get, which is partly because the internet as we know it is ready for disruption if we are to enable a new depth of digital connection.

Additionally, industries are able to reduce errors and increase productivity by using synthetic virtual worlds for training models and robots. Digital twins for factories, cities and other virtual assets can be used to get a product right the first time it is deployed.

Nvidia is a company that is not standing still. Last August, I had predicted Nvidia would surpass Apple to become the world’s most valuable company. The Omniverse is one of many reasons I believe this prediction is still on track to come true.

The I/O Fund is a team of analysts who share their research publicly as they build a portfolio of 20 stocks. Our team has record results for a retail Fund and we also have four-digit gains on some of our free newsletter coverage. You can learn more about our premium service by clicking here or sign up for our free newsletter here.by clicking here or sign up for our free newsletter here.

Disclaimer: This is not financial advice. Please consult with your financial advisor in regards to any stocks you buy.

Throwback: Nvidia will Surpass Apple’s Valuation in 4.5 Years

Posted on January 7, 2022June 30, 2026 by io-fund

Throwback: Nvidia will Surpass Apple’s Valuation in 4.5 Years

Last August, I predicted that Nvidia could surpass Apple on market cap. Here is what I said in my Forbes article: “I believe Nvidia is capable of out-performing all five FAAMG stocks and will surpass even Apple’s valuation in the next five years” and I expanded on this by stating, “We believe [Nvidia] can surpass Apple by capitalizing on the artificial intelligence economy, which will add an estimated $15 trillion to GDP. This is compared to the mobile economy that brought us the majority of the gains in Apple, Google and Facebook, and contributes $4.4 trillion to GDP.”

Source: YCharts

As strong as Nvidia has been on price action, Apple will not allow my prediction to be an easy slam dunk as the heavyweight briefly claimed a $3 trillion market cap.

Source: YCharts

Currently, Nvidia has a market cap of $690 billion and Apple has a market cap of around $2.9 trillion. Nvidia’s market cap rose about 22% compared to Apple’s 17% since my publication of the article. I made this prediction in August of 2021, and during the month of November, we were beginning to make headway with a diversion between semiconductors and big tech.

Source: YCharts

One of the main reasons for me to make the bold statement that Nvidia will surpass Apple’s valuation is that the market opportunity for Nvidia is vast when compared to the mobile economy, which benefitted Apple.

“Artificial intelligence will touch every aspect of both industry and commerce, including consumer, enterprise, and small-to-medium sized businesses, and will do so by disrupting every vertical similar to cloud. To be more specific, AI will be similar to cloud by blazing a path that is defined by lowering costs and increasing productivity.”

When we began covering Nvidia, we were stating the company would become a leader on AI while most analysts were stuck on the gaming storyline as this was Nvidia’s core product for many decades. This caused many investors to miss out on the top supplier for AI accelerator chips in the data center. We had predicted this three years ago when we wrote: Nvidia has two impenetrable moats – which are developer adoption and the GPU-powered cloud. Notice, we did not mention gaming or crypto mining despite this being the only two narratives on this company at the time.

Sign up for I/O Fund's free newsletter with gains of up to 1100% – Click hereSign up for I/O Fund's free newsletter with gains of up to 1100% – Click hereClick here

We published this again in 2019 for premium members when we stated:

Nvidia’s acceleration may happen one or two years earlier as they are the core piece in the stack that is required for the computing power for the front-runners referenced in the graph above. There is a chance Nvidia reflects data center growth as soon as 2020-2021. -published August 2019, Premium I/O Fund.

Since the original 2018 publication on the two impenetrable moats, Nvidia has greatly outperformed FAAMG. We believe the same will be true over the next five years.

Source: YCharts

One reason for this is that last year, Nvidia released the Ampere architecture and A100 GPU as an upgrade from the Volta architecture. The A100 GPUs are able to unify training and inference on a single chip, whereas in the past Nvidia’s GPUs were mainly used for training. This allows Nvidia a competitive advantage by offering both training and inferencing. The result is a 20x performance boost from a multi-instance GPU that allows many GPUs to look like one GPU. The A100 offers the largest leap in performance to date over the past 8 generations.

One year later and the Ampere architecture is becoming one of the best-selling GPU architectures in the company’s history. This quarter, Microsoft Azure recently announced the availability of Azure ND A100 v4 Cloud GPU which is powered by NVIDIA A100 Tensor Core GPUs. The company claims it to be the fastest public cloud supercomputer. The news follows the launch by Amazon Web Services and Google Cloud general availability in prior quarters. The company has been extending its leadership in supercomputing. The latest top 500 list shows that Nvidia power 342 of the world’s top 500 supercomputers, including 70 percent of all new systems and eight of the top 10. This is a remarkable update from the company.

There are many other catalysts that will help Nvidia become the world’s most valuable company to prove my prediction true, including the metaverse, automotive, data analytics such as Spark with GPU acceleration, virtual machines for AI workloads and perhaps edge devices by licensing (or acquiring) Arm architecture.

We only have to wait until August of 2026 to see if Nvidia did indeed pass up Apple’s market cap. However, the wait should be an easy one if Nvidia continues to treat investors to the smooth gains (like butter) we’ve seen as of late.

Disclaimer: This is not financial advice. Please consult with your financial advisor in regards to any stocks you buy.