TSMC released its September monthly revenue, which grew 39.6% YoY and 0.4% MoM to NT$251.87 billion. The third quarter revenue grew 39% YoY to NT$759.69 billion, beating the LSEG estimates of NT$750.36 billion.
In US dollar terms, Q3 revenue grew by 36.2% YoY to $23.53 billion using the average exchange rate of $1=NT$32.28. We will get the official USD figures when the company releases its full results on October 17th. The company handsomely beat the upper end of the guidance of $22.4 billion to $23.2 billion.
The strong September results suggest surging demand for AI chips and with more details on full quarter performance coming this week. Due to its technological leadership, the company has over 90% market share in the manufacturing of advanced AI chips. TSMC's customer base includes leading companies like Apple, Nvidia, AMD, Qualcomm, Marvell, and Intel.
TSMC is the leading global foundry in terms of revenue. It has a market share of 62.3% in Q2 2024. Samsung ranks a distant second with a market share of 11.5% in Q2 2024. WSJ reported TSMC’s global foundry revenue share to hit about 64% in 2024, up from 51% in 2019 and Samsung’s share to drop to 10% from 16% in 2019.
Revenue
Strong revenue growth is expected in the coming quarters due to the robust demand for AI and smartphone chips. Revenue estimates have gone up over the past few quarters, 4.7 points for the current quarter and estimates have risen 7.6 points for Q1 of next year indicating analysts’ confidence in the company’s ability to maintain growth.
Management’s guide for the third quarter is between $22.4 billion and $23.2 billion. This represents year-over-year growth of 31.9% at the midpoint. Based on the monthly TWD figures, Q3 revenue grew by 36.2% YoY to $23.53 billion using the average exchange rate of $1= NT$32.28. We will get the official USD figures when the company releases its full results on October 17th.
Q2 revenue grew by 32.8% YoY and 10.3% QoQ to $20.82 billion, primarily driven by strong AI demand and partially offset by smartphone seasonality. Q2 reported a remarkable 19.9% acceleration from 12.9% growth in Q1.
During the Q2 earnings call, management increased full-year 2024 revenue guidance from low to mid-20% to slightly above mid-20% in US dollar terms as strong AI and high-end smartphone demand will lead to an increased capacity utilization for the 3-nanometer and 5-nanometer process technologies in the second half of the year.
Margins
While many companies struggle with rising costs, TSMC has successfully navigated these challenges by controlling costs and negotiating better prices with its customers. Due to reducing costs during a growth phase, referred to as “economies of scale,” and its leadership position in the foundry industry, the company has been able to report strong profitability.
According to DigiTimes, TSMC has notified its clients that prices for its 3-nanometer and 5-nanometer process products will increase by 3 to 8% in 2025. In the last earnings call, management hinted that prices will increase due to cost escalation.
Management has guided Q3 gross margin to increase 1.3 percentage points sequentially to 54.5% at the mid-point due to better capacity utilization, cost improvements, and productivity gains compared to 54.3% in Q3 last year. The margin is expected to be partially offset by the N3 ramp, N5 to N3 tool conversion costs, and higher electricity prices in Taiwan. Electricity prices in Taiwan increased by 17% last year and another 25% in April this year. Management is confident of achieving a long-term gross margin of 53% and higher.
Operating margins are expanding, helped by operating leverage. Management guide for Q3 is 43.5% at the midpoint compared to 41.7% in Q3 2023.
Net income was $7.66 billion or 36.8% of revenue which is lower than usual. This margin trends in the high 30% range and into the 40% range. Return on equity was 26.7% compared to 23.2% in the same period last year.
EPS
EPS is projected to rise significantly. Analysts expect Q3 GAAP EPS to grow 36.4% YoY to $1.76. For next quarter, EPS is expected to grow 34.7% YoY to $1.94 in Q4.
Compare this to Q2 GAAP EPS of $1.48, up 29.8% YoY, which beat estimates by 4.2% due to better capacity utilization, cost improvement and operating leverage. More importantly, analysts expect Q3 EPS to grow 18.9% sequentially.
Analysts expect 2024 GAAP EPS to grow 26.4% YoY to $6.55.
For FY2025, they expect GAAP EPS to grow 29.9% YoY to $8.51.
Cash Flows and Balance Sheet
The company’s financial stability is evident due to its strong operating cash flows, which have more than doubled in Q2. The foundry industry is capital-intensive and this is why you will notice a wide difference between operating cash flows and free cash flows for the company.
Operating cash flow was $11.68 billion or 56.1% of revenue compared to 35% of revenue in the same period last year and 73.6% in Q1.
Free cash flow was $5.32 billion or 25.5% of revenue compared to (-17%) of revenue in the same period last year and 43% in Q1. The company had negative free cash flows last year due to the income tax payment of $3.85 billion.
Management expects strong AI demand to continue and raised the midpoint of the capex for 2024 to $31 billion from the previous $30 billion, up 1.8% YoY. About 70% to 80% of the capex will be allocated to the advanced process technologies.
According to a report from Economic Daily News, Institutional Investors expect the capex to remain within the updated guide for this year. They also expect management to announce an increase in capex for next year during the January results due to the expected strong demand for 2 nm process technology.
Cash and marketable securities were $63.05 billion and debt of $30.4 billion compared to $60 billion and $30.25 billion in Q1. The company paid $2.8 billion in dividends in Q2.
Revenue by Platform
As the leading foundry for AI accelerators, TSMC is riding the enormous wave of demand from Big Tech. The chipmaker’s high-performance computing (HPC) revenues rose 28% QoQ to $10.8 billion and accounted for 52% of Q2 revenue, up from 46% of revenue in Q1. The HPC segment is above the 50% mark for the first time.
We can also notice in the below chart that the HPC revenue (which are mainly AI-related) reached a record $10.8 billion in Q2.
Smartphone revenues declined (-1%) QoQ due to seasonality and accounted for 33% of revenue compared to 38% of revenue in Q1. In the Q2 earnings call, management mentioned that they are witnessing strong AI and high-end smartphone-related demand in Q3. We will get the exact mix this week.
Internet of Things revenue grew by 6% sequentially and accounted for 6% of revenue.
Automotive revenue increased 5% sequentially and accounted for 5% of revenue. Digital Consumer Electronics increased 20% sequentially accounting for 2% of revenue. Other revenue increased 5% and accounted for 2% of revenue.
Revenue by Technology
The Advanced nodes are defined as 7-nanometer and below. We discussed in our editorial on the advanced nodes and AI-related revenue reaching fresh records. Most of the AI chips produced by the company utilize 5-nanometer and 4-nanometer process technology. 3-nanometer revenue is expected to triple this year and Apple usually gets the preference for the most advanced node in production. Volume production for 2-nanometer is expected in 2025 and should have a meaningful revenue contribution in the first half of 2026.
In Q2 2024, 3-nanometer process technology contributed 15% of wafer revenue, while 5-nanometer and 7-nanometer accounted for 35% and 17%, respectively.
In Q1 2024, 3-nanometer process technology contributed 9% of wafer revenue, while 5-nanometer and 7-nanometer accounted for 37% and 19%, respectively.
Advanced Packaging
The AI wave has also boosted the company’s advanced packaging business, particularly Chip-on-wafer-on-substrate (CoWoS). Morgan Stanley expects CoWoS capacity to reach 80,000 to 90,000 wafers per month by the end of 2025, up from a prior estimate of 70,000. The 2025 production capacity would suggest an over 430% increase from 15,000 at the end of 2023.
Management said in the Q2 earnings call Q&A that supply is expected to continue to be tight next year. They also mentioned that they are working with OSAT (Outsourced Semiconductor Assembly and Test) partners to increase production capacity.
Valuation
The company trades at a P/E ratio of 34.4 and a forward P/E ratio of 29.1. The P/S ratio is 13 and a forward P/S ratio of 11.4. In the last five years, the P/E ratio peaked at 41.8 in February 2021 and hit a low of 10.3 in November 2022. The stock is now trading above its five-year average P/E ratio of 24.1.
Conclusion
Due to technological leadership, TSMC can capture a significant portion of AI business and is the common denominator to the biggest AI winners as it supplies to Nvidia, Apple, AMD, Marvell, and Qualcomm. The company is negotiating better prices with its customers, is making cost improvements, and is maintaining strong margins and cash flows. Since TSMC generates strong cash flows, it has the scale to invest for future growth, and barriers of entry are high as the foundry industry is capital-intensive. We look forward to adding this stock to our portfolio in the coming months and will provide the play-by-play for our entries and adds to Essentials Members.
Pro premium members receive deep-dive research on the stocks in the portfolio and quarterly earnings kickoff webinars. In addition, the Advanced Market Signals Members receive regular technical and broad market analysis and weekly webinars from our Portfolio Manager, Knox Ridley. We have also recently discussed with the Advanced Market Signals members an AI stock that is up 415% in the last three months. Learn more here.here.
Royston Roche, Equity Analyst at the I/O Fund, contributed to this article.
Micron beat estimates on the top and bottom line in Q4, driven by strength in data center DRAM/HBM and a record for NAND. In addition, the company’s Q1 guide far exceeded consensus estimates. Margins expanded sequentially primarily due to HBM and operating cash flow also saw significant sequential growth (whereas free cash flow was much lower due to higher capex).
The report looks to have eased concerns as there were quite a few analyst downgrades following the conferences held in August. We were net buyers in September of Micron, and so we are pleased to see the air cleared on this stock. Analysts had downgraded (and even double downgraded) the stock on comments that bit shipments would be flat to slightly up. The report revealed that despite flat DRAM bit shipments, prices increased in the mid-teens for DRAM, leading to the beat. For NAND, bit shipments increased in the high single-digit percentage and prices increased in the high single-digit percentage range.
Management said they “forecast record revenue in fiscal Q1 and a substantial revenue record with significantly improved profitability in fiscal 2025.” Driven by strong HBM demand, management added that its “mix of data center revenue reached a record level in fiscal 2024, and we expect will grow significantly from here in fiscal 2025.”
Revenue
Micron reported revenue growth of 93% YoY to $7.75 billion in Q4, closing out FY24 with nearly 62% YoY growth to $25.1 billion in revenue. FY24’s recovery was broad-based, with DRAM and NAND both reporting substantial YoY increases.
DRAM revenue increased 14% QoQ to $5.3 billion in Q4, representing 69% of total revenue. For FY24, DRAM revenue totaled $17.6 billion, increasing 60% YoY.
NAND revenue rose 15% QoQ to a record $2.4 billion in Q4, representing 31% of total revenue. For FY24, NAND revenue was $7.2 billion, increasing 72% YoY.
For Q1, management guided for revenue growth of nearly 84% YoY to $8.7 billion at midpoint, easily surpassing the $8.2 billion consensus estimate. Q1’s guide still points to a 900 bp QoQ deceleration with Q4 remaining Micron’s peak growth quarter.
Current estimates for FY25 revenue sit at $37.6 billion for 50% YoY growth, though it’s likely that this gets revised higher in the coming days based on the size of Q1’s guidance beat.
Margins
Micron topped all of its guided figures for margins in the fourth quarter, with gross, operating and net margin all significantly expanding sequentially. For Q1, this expansion is set to continue, with management guiding for operating margin to expand to the mid to high 20% range.
GAAP gross margin in Q4 was 35.3%, ahead of management’s guide for 33.5% and expanding from 26.9% the prior quarter. Adjusted gross margin was 36.5%, expanding from 28.1% last quarter and (9.1%) in the year ago quarter. Management said that in Q4, “HBM remained accretive to both DRAM and overall company gross margins.”
For Q1, management guided GAAP gross margin of 38.5%, implying another sequential expansion of ~320 bp. Adjusted gross margin was guided at 39.5%, Management emphasized that Q1’s gross margin “is projected to improve sequentially primarily on better pricing and portfolio mix.”
GAAP operating margin in Q4 was 19.6%, up 9 percentage points sequentially, and ahead of management’s guide for 17.8%. Adjusted operating margin was 22.5%, up from 13.8% last quarter and ahead of the guide for 20.5%.
For Q1, management’s guided operating expenses implied a GAAP operating margin of 24.6%, another 5 percentage point sequential increase.
Micron closed out FY24 with significant improvement in both gross and operating margin:
FY24 GAAP gross margin was 22.4%, up more than 31 points from (9.1%) in FY23.
FY24 GAAP operating margin was 5.2%, up more than 42 points from (37%) in FY23.
Net margins also significantly improved due to the increased leverage.
For Q4, net margin was 11.4%, up from 4.9% last quarter as net income rose more than 167% sequentially to $887 million. Adjusted net margin was 17.3%, up from 10.3% last quarter.
EPS
GAAP EPS was $0.79 in Q4, up more than 163% QoQ from $0.30 in Q3 and a substantial improvement from a loss of ($1.31) in the year ago quarter. Adjusted EPS was $1.18, up 90% QoQ from $0.62 in Q3 and again a significant improvement from ($1.01) in the year ago quarter. Management said this was “driven by better pricing and profitability.”
What these numbers reflect is that Micron has officially bottomed on EPS and is firmly returning to positive growth.
Micron guided for sequential growth to continue, expecting Q1’s GAAP EPS to be $1.54, up nearly 95% QoQ. Adjusted EPS was guided at $1.74, well ahead of estimates for $1.59, and it’s likely to push Q2 and Q3 estimates higher as well.
Cash Flows and Balance Sheet
Micron grew operating cash flow by nearly $1 billion QoQ; however, increased capex kept adjusted FCF growth to a minimum.
Q4’s operating cash flow was $3.41 billion, or 44% of revenue, up from $249 million or 6.2% of revenue in the year ago quarter. For FY24, operating cash flow totaled $8.51 billion, increasing more than 445% YoY from $1.56 billion in FY23.
Adjusted FCF in Q4 was just $323 million, or 4.2% of revenue, compared to $425 million in Q3, as Micron spent ~$3.1 billion on capex in the quarter. For FY24, adjusted FCF was minimal, at $386 million, or 1.5% of revenue, though this was a stark contrast to FY23’s adjusted FCF of ($5.44 billion), or (37.1%) of revenue.
It was stated capex would be meaningfully higher in fiscal 2025 at mid-30s percentage range to support “growth in both greenfield fab construction and HBM” investments, as Micron works to build out its fabs in New York and Idaho in FY25 and FY26. Capex totaled $8.1 billion in FY24, but management expects FY25 capex to be “meaningfully higher and at around the mid-30s percentage range of revenue,” suggesting capex easily surpasses $10 billion next year.
The company stated wafer capacity is below peak levels, partly due to an increasing mix of HBM that is reducing DRAM supply for traditional products. The capex spending is needed to continue to supply HBM. There is also a low-capex environment for NAND at the moment, and it was stated this would ultimately lead to healthy NAND supply-demand dynamics.
Inventory was $8.9 billion, or 158 days, and Micron expects to draw down this inventory to support revenue growth in FY25.
Cash, equivalents and investments totaled $9.16 billion, while debt totaled $13.4 billion.
Business Units
Compute and Networking (CNBU) revenue was $3.02 billion, up 14% QoQ and 152% YoY. This was significant growth acceleration, up from 85% YoY in Q3 and 59% YoY in Q2.
Management said that “data center server DRAM achieved a quarterly revenue record in fiscal Q4, driven by strong demand for high-capacity solutions as well as our continued ramp of HBM.” The company expects HBM TAM to grow from $4 billion in CY23 to over $25 billion in CY25. As a percent of overall DRAM, HBM will grow from 1.5% in 2023 to 6% in 2025. Micron reiterated it will be able to capture a similar market share of HBM as it has in DRAM, which was 21.5% of market share in early 2024.
Mobile (MBU) revenue increased 18% QoQ to $1.88 billion, though YoY growth of 55% decelerated from 94% YoY in Q3. Management said the growth was driven by seasonal product launches.
Micron provided a hint as to when investors can expect to see AI PC growth, which looks to be H2 2025: “PC unit volumes remain on track to grow in the low single-digit range for calendar 2024. We expect unit growth to continue in 2025 and accelerate into the second half of calendar 2025 as the PC replacement cycle gathers momentum with the rollout of next-gen AI PCs, end of support for Windows 10 and the launch of Windows 12.”
The timing was repeated again: “So that too plays a factor. And of course, I would just like to remind you that we have pointed out that overall smartphone and PC, unit growth will be occurring in 2025 and of course, increasing penetration of AI phones and second half, that acceleration, that growth will be obviously stronger than the first half.”
On average, PCs require 12GB of DRAM today with AI PCs needing a minimum of 16GB and up to 32 to 64GB of DRAM. We covered this previously here. Micron’s LP series for PCs offer are low-power DRAM modules that provide 60% lower power and up to 70% better performance with 60% space savings.
Mobile devices require 8GB of DRAM whereas AI-powered mobile devices will come with 12GB to 16GB of DRAM.
Storage (SBU) revenue rose 24% QoQ and 127% YoY to $1.68 billion, with the YoY growth rate accelerating from 116% in Q3. Management said the growth was “led by data center SSD, which reached a quarterly revenue record,” while NAND storage reached a record for the full year.
Embedded (EBU) was the only segment to record a sequential decline in Q4, with growth falling (9%) QoQ but rising 36% YoY to $1.17 billion. Management added that the “automotive segment achieved a new fiscal year revenue record for the fourth consecutive year.”
Earnings Call and Addt’l Commentary
The company stated they are upgrading their expectation for calendar 2024 industry DRAM bit demand growth to be in the high-teens percentage range. It was further stated: “In calendar 2025, we expect both DRAM and NAND industry bit demand growth to be around the mid-teens percentage range.”
The company also stated: “We see increasing DRAM and NAND content both in traditional as well as AI servers” and that “our mix of data center revenue reached a record level in fiscal 2024 and we expect will grow significantly from here in fiscal 2025.”
There is a slight slowdown in management’s guide for DRAM for next year, as it’s being stated as growth in the high teens is expected for 2024 while growth in the low teens is expected for 2025. As we noted in our pre-earnings writeup, the slowdown is coming from AI PCs and smartphones.
“At 2024, we have increased the outlook to high teens based on the strength in the data center. And 2025, as we look at it, just keep, in fact — mind two factors: one is we are now comparing it to the higher base of 2024, which has gone to high teens. So that, of course, impacts the percentage of the '25. And second piece is that, as we have noted, that smartphone and PC, which at the end market level are continuing to do fine.
But given for the 3 factors that we have mentioned in our earnings call script that the customers built some inventory. The sell-in is somewhat less than their sellout in terms of memory, and we have said that by spring of 2025, we expect in PCs customer inventory levels to get to healthier levels versus now, and these will continue to improve.”
Another factor is that HBM3e is leading to wafer capacity constraints. It has a 3:1 trade ratio, which which means it takes 3X more wafers to produce HBM3e.
High Bandwidth Memory (HBM)
We’ve covered the importance of HBM for some time now, which you can reference here.
This quarter, Micron started shipments of HBM3e which are 12 high, 36-gigabyte units. These units provide up to 20% lower power consumption and 50% higher DRAM capacity than its competitors’ 8 high, 24 gigabyte solutions. As Micron stated during the call when asked about competitors, the company’s strategy is to provide the best HBM on “planet earth.” Micron will continue to increase its mix of HBM3e 12 in early CY2025.
As GPUs move toward a 1-year road map, so will Micron. HBM4 will be shipping in 2026.
Last quarter, the stock sold off following commentary that Micron’s HBM is contracted 6-8 quarters out. The company reiterated this again, yet it’s clear the exact contract pricing details are not being disclosed. This was evident in the beat in today’s report.
This was an important statement regarding next year’s setup:
“And we are looking at strong momentum, not just with HBM. We have talked about multiple billions of dollars of revenue that we target to generate in our fiscal year 2025 from high-capacity DRAM modules as well as LP memory in data center, so these are all the elements that point to strong demand trends and demand trends driven by AI in data center as well as in smartphone and PCs where more and more content is required in an environment where the leading-edge supply is today tight.
So I think the opportunity is tremendous, and we see healthy demand supply balance and a constructive environment for our financial performance in fiscal 2025. And that's why we say with confidence that we'll deliver a substantial revenue record in fiscal year 2025, the significant improvement in our profitability as well.”
DDR5 SDRAM memory products reduce power consumption while doubling bandwidth. The lower power DRAM is assisted with LP5 solutions that increase speed while lower power requirements. The LPDDR designs were originally designed for mobile yet have been optimized for AI server infrastructure.
Data Center Storage up 300% in One Year
Micron’s strategy for vertical integration of SSDs has resulted in Micron seeing a “quarterly revenue record” of over $1 billion in revenue and data center SSDs in fiscal Q4. The company stated “our fiscal 2024 data center SSD revenues more than tripled from a year ago.”
Conference Commentary
As we covered in our pre-earnings report, conferences in August led to many analysts downgrading Micron. The first question on the call was about the confusion this caused.
Timothy ArcuriUBS Investment Bank
Mark, I guess my first question is, some of the assumptions in guidance. I think you've been saying kind of on the conference circuit that bits would be pretty flat in fiscal Q1 for both DRAM and NAND. Is that what you're still assuming so that most of the increase in the revenue is basically pricing? Is that correct?
Mark MurphyExecutive VP & CFO
Tim, what we see now and we had provided a slight update in August, but we now see that DRAM bits, we expect to be up somewhat higher than what we had said before. We had said before they were going to be flat and then we revised that to flat to slightly up, and in this latest guide, we now view DRAM to be up somewhat higher from that. NAND bits, we expect to be sequentially flattish.”
Conclusion:
Micron had a fantastic earnings report and is sitting at an attractive valuation, which cannot be said for many AI stocks right now. The company has been doubted by Wall Street for most of the year, and yet this report easily places Micron as a top choice among AI peers as we head into 2025.
It makes sense Wall Street would doubt Micron in CY2024 as the company had several millions in HBM3 revenue this year, with is low compared to some of the heavyweights. Yet, the AI-related portion of its revenue is not only going to ramp next year, but it’s also accretive to margins. It’s this last piece that is key as AI markets such as custom silicon or AI servers have the opposite effect and weigh on margins. Keep this in mind as we move along.
We are glad to have Micron in our portfolio and are looking forward to building a bigger position in time.
Broadcom boosted AI revenue outlook for the year from $11 billion to $12 billion. Based on the forecasted $51.5 billion in revenue in fiscal 2024, AI is expected to contribute more than 23% of Broadcom’s revenue this year.
During the CEO’s opening remarks, more information about Q3’s AI revenue and Q4’s AI revenue was provided: “AI demand remains strong and we expect, in Q4, AI revenue to grow sequentially 10% to over $3.5 billion. This will translate to AI revenue of $12 billion for fiscal '24, up from our prior guidance of over $11 billion.” The issue is that AI revenue was flat QoQ compared to being up 35% QoQ last quarter.
Overall, Q3 did beat yet Q4’s guide was a bit light, with Broadcom forecasting revenue of $14 billion next quarter versus consensus for $14.04 billion. Semiconductor solutions revenue growth also missed growth forecasts with YoY growth of 5%, while estimates from Oppenheimer expected 7% growth.
The stock is priced to perfection, and we do not think this earnings report is not enough to sustain the valuation. We discuss this and more below.
Revenue
Broadcom beat Q3’s revenue estimates marginally and was one percent ahead of consensus estimates, compared to a nearly 4% beat last quarter. Q4 was also guided slightly below analyst estimates.
Q3 revenue was $13.07 billion for YoY growth of 47.3%, accelerating 430 bp from 43% YoY growth in Q2. Excluding VMWare’s contribution, YoY growth was 4%.
For Q4, management guided for $14 billion in revenue, for YoY growth of 50.6%, marking another 330 bp acceleration, but short of the 51% growth estimate from analysts.
For FY24, management did not provide an updated revenue guide, with the previous guidance provided of $51 billion. However, given Q4’s guide, revenue for FY24 is projected to be $51.5 billion.
As probably the most-closely watched (and most anticipated) figure, AI revenue for FQ3 was $3.1 billion, flat QoQ. Through Q3, AI revenue totaled $8.5 billion. For fiscal 2024, Broadcom increased its AI revenue outlook to $12 billion, a 20% increase from $10 billion in Q1 and another $1 billion increase from last quarter’s view for $11 billion. The increased outlook is driven by strong growth for Ethernet and ASICs, and implies Q4 AI revenue of $3.5 billion, up nearly 13% QoQ.
The issue that remains is, will this AI revenue be enough to justify the valuation (in the near term)?
Key Segments
Semiconductor solutions revenue was $7.27 billion, increasing 5% YoY and 1% QoQ. This was a 100 bp deceleration from 6% YoY growth last quarter, while also falling short of estimates, with some analysts expecting growth of ~7% YoY and 2% QoQ.
Networking grew 43% YoY to $4 billion. Ethernet switching, driven by Tomahawk 5 and Jericho3-AI grew over 4x year-on-year. We’ve covered these products here. Within networking, lasers and dies used in optical interconnects grew 3X, PCI Express switches grew 2X and Broadcom is shipping 5nm 400GB NICs and 800GB DSPs.
The CEO stated non-AI networking bottomed in Q2 and was up 17% QoQ yet down (-41%) YoY
The CEO stated custom AI accelerators grew 3.5X year-over-year. Notably, we did not get a QoQ number which is where the weak metric was at $3.1B AI revenue flat QoQ.
Infrastructure solutions revenue was $5.80 billion, accelerating to 200% YoY growth from 175% YoY in the prior quarter. The segment was up 9.7% QoQ and this segment also grew QoQ last quarter, which helps to illustrate the acceleration from VMWare.
VMWare contributed $3.8 billion with the acquisition helping to drive the strong YoY growth.
VMWare’s annualized booking value (ABV) was up 32% QoQ to $2.5 billion, and is shaping up to be a major piece to Broadcom’s story. You can read more about this here. Two quarters ago, the ABV for VMWare was $1.2 billion, proving there has been a sudden acceleration underway. The VMWare segment is expected to be quite profitable, achieving adjusted EBITDA of $8.5 billion by next fiscal year. During the Q&A portion, it was discussed the overall software margin will remain between 80% and 90%.
Aligned with the commentary that the non-AI segments have bottomed (with the exception of Broadband), the remaining revenue segments reported QoQ growth despite being down double digits YoY.
Server storage connectivity revenue was $861 million, up 5% QoQ yet down (-25%) YoY. Server storage is expected to grow QoQ yet will be down YoY
Wireless revenue of $1.7 billion grew 1% YoY and is expected to grow 20% QoQ next quarter.
Broadband revenue declined 49% YoY and is expected to bottom in the beginning of 2025.
On the call, an analyst asked if these segments will return to prior levels, to which the CEO stated they would and he cited bookings as an indication the bottom is likely behind them.
Hock Tan:
“As you all know, we've gone through your typical down cycle of semiconductors. And I'm referring particularly to non-AI, and we have talked about that before many times. We've gone through a down cycle as the ecosystem, as many of our customers, but the broad ecosystems, work on an adjustment in inventory levels in all stages in the supply chain. And we're not immune from it, obviously as we try to insulate ourselves from it as much as possible. We've gone through it. And the signs on the indications we have seen very clearly is we have, in fact, passed through the bottom. The best indicator is the bookings we are receiving. In non-AI, our bookings in Q3 of non-AI semiconductor demand is up 20%. And so that tells us we are well on the way to recovery.”
Margins
As outlined in our pre-earnings analysis, the VMWare merger integration has weighed on margins so far in fiscal 2024, though Q3 showed more positive signs on margin recoveries. Adjusted margins strengthened sequentially across the board, with larger growth visible down the line.
Gross margin was 63.9% in Q3, down from 69.5% in the year ago quarter but up from 62.3% in Q2. Adjusted gross margin was 77.4%, up from 75.1% last year and 76.2% last quarter.
Operating margin was 29.0% in Q3, up from 23.7% last quarter and a remarkable increase from 17.3% in Q1; however, Q3’s margin remained much lower on a YoY basis, down from 43.4% last year. Adjusted operating margin was 60.8%, up from 57.2% last quarter but decreasing from 62.4% last year.
Net margin was (14.4%) for Q3, due to a one-time non-cash tax provision of $4.5 billion in the quarter. Adjusted net margin was 46.8%, up from 43.2% last quarter but down from 51.8% last year.
Regarding the tax liability, there was a question in the Q&A session if it was related to selling assets, to which the CFO responded that it was not due to a sale of assets, rather: “It's just we relocated the IP and that caused the $4 billion charge. The offset to that is a deferred tax liability. So think of that as noncash, very little cash impact to that.”
EPS
GAAP EPS was ($0.40), not comparable to estimates for $0.55 due to the $4.5 billion tax provision in the quarter. Adjusted EPS was $1.24, beating estimates by $0.04, and representing YoY growth of 18% and QoQ growth of nearly 13%.
Adjusted EPS growth is currently estimated to accelerate to nearly 24% YoY in Q4, and to the low-30% range for the first half of fiscal 2025. Given the sequential rebound in margins (partially due to better controlled costs at VMWare) and adjusted EPS beat in Q3, these growth projects may get revised higher in the coming days.
Q3’s adjusted EBITDA was 62.9%, with growth of 42% YoY to $8.22 billion. This was a solid improvement from 59.5% in Q2. Management guided for a 64% adjusted EBITDA margin in Q4, with the sequential improvement being driven by the integration of VMWare as Broadcom charts a path to pre-acquisition EBITDA margins.
Cash Flows and Balance Sheet
Cash flows and cash flow margins improved sequentially, while total debt took a step lower this quarter.
Operating cash flow was $4.96 billion, increasing more than 5% YoY and 8% QoQ. Operating cash flow margin as 38.0%, improving from 36.7% last quarter, though this is lower than pre-acquisition margins of 53.2% in the year ago quarter.
Free cash flow was $4.79 billion, rising more than 4% YoY and nearly 8% QoQ. Free cash flow margin was 36.7%, increasing from 35.6% last quarter but also lower than the 51.8% margin in the year ago quarter.
Cash, equivalents and investments totaled $9.95 billion.
Total debt was $69.96 billion at the end of Q3, down from $74.02 billion in Q2, with Broadcom repaying more than $9.2 billion of debt in the quarter, offset by nearly $5 billion in proceeds from long-term borrowings.
Valuation:
Broadcom is priced for perfection at 32 forward PE Ratio and 14 forward PS Ratio. We’ve covered semis since 2018 and a quality semiconductor stock rarely trades at these levels, it’s normally in the 6-8 forward PS range.
The 3-year median is only available on current PE Ratio, but I have a 3-year PE ratio of 28.5 compared to the current PE ratio of 65.8. The 5-year median is at 39.5.
Earnings Call:
Analysts asked in various ways what growth rate they could expect on AI revenue moving forward, but to no avail on any specific numbers. Here was the first question on the call:
Question Vivek Arya (Analysts)
Just a clarification, Hock, and then the question. So I think AI revenue roughly $3.1-ish billion in Q3, flattish sequentially. What was the mix in terms of compute versus networking? And the $3.5 billion for Q4, what do you see of that mix? And then as we get into fiscal '25, I realize you're not guiding overall AI, but just how is your general kind of confidence and visibility? Do you think that Broadcom can kind of grow in line or better than the overall AI silicon industry in fiscal '25?
Answer Kirsten Spears (Executives)
Yes. Well, as we indicated in the last earnings call, for this past quarter, I think we're talking about 2/3 in compute and 1/3 in networking. And we kind of expect Q4 to run the similar trend. And to answer your second part, no, we don't guide yet for fiscal '25, but we do expect fiscal '25 to continue to be strong, to show strong growth on our AI revenue.
-End Quote
There was a more tempered tone in terms of timing. For example, when asked where custom silicon is in the adoption curve, it was stated it will take some time.
Question Edward Snyder (Analysts)
Right. That basically suggests that you're on the early part of your curve where I'm not trying to call the GPUs whatever, but you could be getting to something closer to the peak of the GPU market just because everything, right, beside the cost expense and as you're spending all this money and you're paying all this money for power, the ASICs become more and more attractive. So the curves are going to look different, right?
Answer Hock Tan (Executives)
It's an accelerating curve. It may take longer than we all want it to happen but definitely accelerating because the size of the demand from those hyperscalers will totally rival that in the enterprise.
-End Quote
Conclusion:
Given AI growth has plateaued this quarter from its rapid growth, there will likely be a re-rating of Broadcom’s valuation sometime in the next 1-2 quarters. Our goal is to trim this position and buy back at lower levels. This strategy requires remaining steadfast to the bigger picture as this company is setting up to be a clear winner over the next decade. We want to do our best at actively managing the position while not losing sight of the bigger picture. With that said, valuation is a common pitfall for tech investors, who grow complacent with their positions. Our firm works to avoid valuation traps, and thus we plan to follow our disciplined process, which is to actively manage stocks that are richly valued.
Damien Robbins, I/O Fund Equity Analyst, contributed to this analysis
Broadcom boosted AI revenue outlook for the year from $11 billion to $12 billion. Based on the forecasted $51.5 billion in revenue in fiscal 2024, AI is expected to contribute more than 23% of Broadcom’s revenue this year.
During the CEO’s opening remarks, more information about Q3’s AI revenue and Q4’s AI revenue was provided: “AI demand remains strong and we expect, in Q4, AI revenue to grow sequentially 10% to over $3.5 billion. This will translate to AI revenue of $12 billion for fiscal '24, up from our prior guidance of over $11 billion.” The issue is that AI revenue was flat QoQ compared to being up 35% QoQ last quarter.
Overall, Q3 did beat yet Q4’s guide was a bit light, with Broadcom forecasting revenue of $14 billion next quarter versus consensus for $14.04 billion. Semiconductor solutions revenue growth also missed growth forecasts with YoY growth of 5%, while estimates from Oppenheimer expected 7% growth.
The stock is priced to perfection, and we do not think this earnings report is not enough to sustain the valuation. We discuss this and more below.
Revenue
Broadcom beat Q3’s revenue estimates marginally and was one percent ahead of consensus estimates, compared to a nearly 4% beat last quarter. Q4 was also guided slightly below analyst estimates.
Q3 revenue was $13.07 billion for YoY growth of 47.3%, accelerating 430 bp from 43% YoY growth in Q2. Excluding VMWare’s contribution, YoY growth was 4%.
For Q4, management guided for $14 billion in revenue, for YoY growth of 50.6%, marking another 330 bp acceleration, but short of the 51% growth estimate from analysts.
For FY24, management did not provide an updated revenue guide, with the previous guidance provided of $51 billion. However, given Q4’s guide, revenue for FY24 is projected to be $51.5 billion.
As probably the most-closely watched (and most anticipated) figure, AI revenue for FQ3 was $3.1 billion, flat QoQ. Through Q3, AI revenue totaled $8.5 billion. For fiscal 2024, Broadcom increased its AI revenue outlook to $12 billion, a 20% increase from $10 billion in Q1 and another $1 billion increase from last quarter’s view for $11 billion. The increased outlook is driven by strong growth for Ethernet and ASICs, and implies Q4 AI revenue of $3.5 billion, up nearly 13% QoQ.
The issue that remains is, will this AI revenue be enough to justify the valuation (in the near term)?
Key Segments
Semiconductor solutions revenue was $7.27 billion, increasing 5% YoY and 1% QoQ. This was a 100 bp deceleration from 6% YoY growth last quarter, while also falling short of estimates, with some analysts expecting growth of ~7% YoY and 2% QoQ.
Networking grew 43% YoY to $4 billion. Ethernet switching, driven by Tomahawk 5 and Jericho3-AI grew over 4x year-on-year. We’ve covered these products here. Within networking, lasers and dies used in optical interconnects grew 3X, PCI Express switches grew 2X and Broadcom is shipping 5nm 400GB NICs and 800GB DSPs.
The CEO stated non-AI networking bottomed in Q2 and was up 17% QoQ yet down (-41%) YoY
The CEO stated custom AI accelerators grew 3.5X year-over-year. Notably, we did not get a QoQ number which is where the weak metric was at $3.1B AI revenue flat QoQ.
Infrastructure solutions revenue was $5.80 billion, accelerating to 200% YoY growth from 175% YoY in the prior quarter. The segment was up 9.7% QoQ and this segment also grew QoQ last quarter, which helps to illustrate the acceleration from VMWare.
VMWare contributed $3.8 billion with the acquisition helping to drive the strong YoY growth.
VMWare’s annualized booking value (ABV) was up 32% QoQ to $2.5 billion, and is shaping up to be a major piece to Broadcom’s story. You can read more about this here. Two quarters ago, the ABV for VMWare was $1.2 billion, proving there has been a sudden acceleration underway. The VMWare segment is expected to be quite profitable, achieving adjusted EBITDA of $8.5 billion by next fiscal year. During the Q&A portion, it was discussed the overall software margin will remain between 80% and 90%.
Aligned with the commentary that the non-AI segments have bottomed (with the exception of Broadband), the remaining revenue segments reported QoQ growth despite being down double digits YoY.
Server storage connectivity revenue was $861 million, up 5% QoQ yet down (-25%) YoY. Server storage is expected to grow QoQ yet will be down YoY
Wireless revenue of $1.7 billion grew 1% YoY and is expected to grow 20% QoQ next quarter.
Broadband revenue declined 49% YoY and is expected to bottom in the beginning of 2025.
On the call, an analyst asked if these segments will return to prior levels, to which the CEO stated they would and he cited bookings as an indication the bottom is likely behind them.
Hock Tan:
“As you all know, we've gone through your typical down cycle of semiconductors. And I'm referring particularly to non-AI, and we have talked about that before many times. We've gone through a down cycle as the ecosystem, as many of our customers, but the broad ecosystems, work on an adjustment in inventory levels in all stages in the supply chain. And we're not immune from it, obviously as we try to insulate ourselves from it as much as possible. We've gone through it. And the signs on the indications we have seen very clearly is we have, in fact, passed through the bottom. The best indicator is the bookings we are receiving. In non-AI, our bookings in Q3 of non-AI semiconductor demand is up 20%. And so that tells us we are well on the way to recovery.”
Margins
As outlined in our pre-earnings analysis, the VMWare merger integration has weighed on margins so far in fiscal 2024, though Q3 showed more positive signs on margin recoveries. Adjusted margins strengthened sequentially across the board, with larger growth visible down the line.
Gross margin was 63.9% in Q3, down from 69.5% in the year ago quarter but up from 62.3% in Q2. Adjusted gross margin was 77.4%, up from 75.1% last year and 76.2% last quarter.
Operating margin was 29.0% in Q3, up from 23.7% last quarter and a remarkable increase from 17.3% in Q1; however, Q3’s margin remained much lower on a YoY basis, down from 43.4% last year. Adjusted operating margin was 60.8%, up from 57.2% last quarter but decreasing from 62.4% last year.
Net margin was (14.4%) for Q3, due to a one-time non-cash tax provision of $4.5 billion in the quarter. Adjusted net margin was 46.8%, up from 43.2% last quarter but down from 51.8% last year.
Regarding the tax liability, there was a question in the Q&A session if it was related to selling assets, to which the CFO responded that it was not due to a sale of assets, rather: “It's just we relocated the IP and that caused the $4 billion charge. The offset to that is a deferred tax liability. So think of that as noncash, very little cash impact to that.”
EPS
GAAP EPS was ($0.40), not comparable to estimates for $0.55 due to the $4.5 billion tax provision in the quarter. Adjusted EPS was $1.24, beating estimates by $0.04, and representing YoY growth of 18% and QoQ growth of nearly 13%.
Adjusted EPS growth is currently estimated to accelerate to nearly 24% YoY in Q4, and to the low-30% range for the first half of fiscal 2025. Given the sequential rebound in margins (partially due to better controlled costs at VMWare) and adjusted EPS beat in Q3, these growth projects may get revised higher in the coming days.
Q3’s adjusted EBITDA was 62.9%, with growth of 42% YoY to $8.22 billion. This was a solid improvement from 59.5% in Q2. Management guided for a 64% adjusted EBITDA margin in Q4, with the sequential improvement being driven by the integration of VMWare as Broadcom charts a path to pre-acquisition EBITDA margins.
Cash Flows and Balance Sheet
Cash flows and cash flow margins improved sequentially, while total debt took a step lower this quarter.
Operating cash flow was $4.96 billion, increasing more than 5% YoY and 8% QoQ. Operating cash flow margin as 38.0%, improving from 36.7% last quarter, though this is lower than pre-acquisition margins of 53.2% in the year ago quarter.
Free cash flow was $4.79 billion, rising more than 4% YoY and nearly 8% QoQ. Free cash flow margin was 36.7%, increasing from 35.6% last quarter but also lower than the 51.8% margin in the year ago quarter.
Cash, equivalents and investments totaled $9.95 billion.
Total debt was $69.96 billion at the end of Q3, down from $74.02 billion in Q2, with Broadcom repaying more than $9.2 billion of debt in the quarter, offset by nearly $5 billion in proceeds from long-term borrowings.
Valuation:
Broadcom is priced for perfection at 32 forward PE Ratio and 14 forward PS Ratio. We’ve covered semis since 2018 and a quality semiconductor stock rarely trades at these levels, it’s normally in the 6-8 forward PS range.
The 3-year median is only available on current PE Ratio, but I have a 3-year PE ratio of 28.5 compared to the current PE ratio of 65.8. The 5-year median is at 39.5.
Earnings Call:
Analysts asked in various ways what growth rate they could expect on AI revenue moving forward, but to no avail on any specific numbers. Here was the first question on the call:
Question Vivek Arya (Analysts)
Just a clarification, Hock, and then the question. So I think AI revenue roughly $3.1-ish billion in Q3, flattish sequentially. What was the mix in terms of compute versus networking? And the $3.5 billion for Q4, what do you see of that mix? And then as we get into fiscal '25, I realize you're not guiding overall AI, but just how is your general kind of confidence and visibility? Do you think that Broadcom can kind of grow in line or better than the overall AI silicon industry in fiscal '25?
Answer Kirsten Spears (Executives)
Yes. Well, as we indicated in the last earnings call, for this past quarter, I think we're talking about 2/3 in compute and 1/3 in networking. And we kind of expect Q4 to run the similar trend. And to answer your second part, no, we don't guide yet for fiscal '25, but we do expect fiscal '25 to continue to be strong, to show strong growth on our AI revenue.
-End Quote
There was a more tempered tone in terms of timing. For example, when asked where custom silicon is in the adoption curve, it was stated it will take some time.
Question Edward Snyder (Analysts)
Right. That basically suggests that you're on the early part of your curve where I'm not trying to call the GPUs whatever, but you could be getting to something closer to the peak of the GPU market just because everything, right, beside the cost expense and as you're spending all this money and you're paying all this money for power, the ASICs become more and more attractive. So the curves are going to look different, right?
Answer Hock Tan (Executives)
It's an accelerating curve. It may take longer than we all want it to happen but definitely accelerating because the size of the demand from those hyperscalers will totally rival that in the enterprise.
-End Quote
Conclusion:
Given AI growth has plateaued this quarter from its rapid growth, there will likely be a re-rating of Broadcom’s valuation sometime in the next 1-2 quarters. Our goal is to trim this position and buy back at lower levels. This strategy requires remaining steadfast to the bigger picture as this company is setting up to be a clear winner over the next decade. We want to do our best at actively managing the position while not losing sight of the bigger picture. With that said, valuation is a common pitfall for tech investors, who grow complacent with their positions. Our firm works to avoid valuation traps, and thus we plan to follow our disciplined process, which is to actively manage stocks that are richly valued.
Damien Robbins, I/O Fund Equity Analyst, contributed to this analysis
Broadcom will release its Q3 FY2024 results after market close today. Investors will be closely watching the AI revenue updates for the ASIC market leader. Broadcom has the potential to grab the attention after Nvidia due to the company’s unique position in the AI Infrastructure space. Recently Citi Analyst also highlighted that the company is catching up on Nvidia as the top holding as it adds more AI customers and accretion from VMware. In addition, it could benefit from investor fatigue with Nvidia, which aligns with our thoughts that we have highlighted here.
The main highlight in the last earnings call was the AI revenue FY2024 guide increase from over $10 billion to over $11 billion. The company beat the top-line and bottom-line estimates and also raised the full-year revenue guidance to $51 billion from the earlier $50 billion. Analysts believe the guide was conservative due to the company ramping up new ASIC customers, the potential bottoming of non-AI networking markets, and VMware integration progressing well. The company also raised the FY2024 adjusted EBITDA guide from 60% to 61%, which is positive as margin recovery post-VMware integration is also important for the stock momentum to grind higher.
Revenue
FQ2 revenue grew by 43% YoY to $12.49 billion, up from 34.2% growth in FQ1. Next quarter, revenue is going to accelerate to 46% YoY growth to $12.96 billion and further accelerate to 51% growth in FQ4.
FQ2 was the first full quarter with a contribution from VMware, and organically, it grew by 12%. Revenue beat estimates by 4.0% and was primarily helped by the strong 280% YoY growth in AI revenue of $3.1 billion, offset by cyclical weakness in enterprises and telcos. Management hinted that non-AI semiconductor revenue had bottomed out in Q2 and was likely to have a modest recovery in the second half of the year.
Management increased the FY2024 revenue guide from $50 billion to $51 billion Analysts expect FY2024 revenue to grow 43.8% YoY to $51.51 billion, organic growth is about 10%.
Analysts expect FY2025 revenue to grow 16.8% YoY to $60.15 billion and FY2026 revenue to grow 11.6% to $67.13 billion.
Ji Yoo, Head of Investor Relations, said in the earnings call, “For fiscal '24, we expect revenue from AI to be much stronger at over $11 billion. Non-AI semiconductor revenue has bottomed in Q2 and is likely to recover modestly for the second half of fiscal '24.
On infrastructure software, we're making very strong progress in integrating VMware and accelerating its growth. Pulling all these three key factors together, we are raising our fiscal '24 revenue guidance to $51 billion.”
Margins
The merger integration process will initially drag the margins in FY2024 due to transition costs and VMware’s lower margin profile. However, cost cutting and merger synergies are anticipated to improve margins in the long term.
The FQ2 gross margin was 62.3%, down from 70% in the same period last year and up from 61.7% in the previous quarter. Adjusted gross margin was 76.2%, down from 75.6% last year and up from 75.4% in the previous quarter.
Operating margin was 23.7%, down from 45.9% last year and up from 17.4% in the previous quarter. The operating margin was mainly lower from last year due to the increase in amortization of acquisition-related intangible assets, restructuring charges, and stock-based compensation.
The adjusted operating margin was 57.2%, down from 62% last year and up marginally from 57.1% in the previous quarter. Excluding transition costs, the adjusted operating margin was 59% and remained the same as in the previous quarter.
Net income was $2.1 billion or 17% of revenue compared to $3.48 billion or 39.9% of revenue in the same period last year. The lower net income was mainly due to the points discussed in the above paragraphs and higher interest expenses this year. The adjusted net income was $5.39 billion or 43.2% of revenue compared to $4.49 billion or 51.4% of revenue last year.
EPS
GAAP EPS was $0.44 compared to $0.82 in the same period last year. The adjusted EPS grew by 6.2% YoY to $1.096 and beat estimates by 1.1%, helped by cost savings. The company has been able to reduce VMware spending to $1.6 billion from the previous $2.3 billion pre-acquisition. Management expects to exit Q4 with a spending of $1.3 billion run rate, better than the previous plan of $1.4 billion. It is further expected to stabilize at $1.2 billion post-integration.
Analysts expect adjusted EPS to accelerate to 14.3% YoY growth to $1.20 in FQ3 and to 23.7% YoY growth to $1.37 in FQ4.
Analysts expect FY2024 adjusted EPS to grow 12.4% YoY to $4.75 and accelerate to 27.7% growth to $6.06 in FY2025.
FQ2 adjusted EBITDA was 59.5%, compared to 65.1% in the same period last year and 59.8% in the previous quarter. The drop is mainly due to VMware's lower margin. The post-integration is progressing well, and management also raised the FY2024 adjusted EBITDA guide from 60% to 61%. They also expect VMware's adjusted operating margin to match Broadcom’s software margin by FY2025.
Cash Flow and Balance Sheet
The company has high debt as it has been growing through successful acquisitions. While high debt is a concern, the company is focusing on repaying medium-term debt and has strong cash flows. Also, the company’s debt prior to the VMware acquisition has long maturities.
Operating cash flow was $4.58 billion or 36.7% of revenue compared to $4.5 billion or 51.6% of revenue in the same period last year and 40.3% in the previous quarter.
Free cash flow was $4.45 billion or 35.6% of revenue compared to $4.38 billion or 50.2% of revenue last year and 39.2% in the previous quarter. Free cash flow excluding cash used for restructuring and integration was $5.3 billion or 42% of revenue. Free cash flow as a percentage of revenue declined from last year due to higher interest expenses related to debt for VMware acquisition and “higher cash taxes due to a higher mix of US income and the delay in the reenactment of Section 174.”
Cash was $9.81 billion and debt of $74.02 billion compared to $11.9 billion and $75.9 billion in the previous quarter. The weighted average coupon rate and term to maturity of $48 billion fixed rate debt is 3.5% and 8.2 years, respectively. The weighted average coupon rate and term to maturity of floating rate debt are 6.6% and 2.8 years, respectively. The company repaid $2 billion of floating rate debt in FQ2 and plans to maintain this quarterly repayment throughout FY2024.
The company paid $2.4 billion in dividends and $1.5 billion in withholding taxes due to the vesting of employee equity, eliminating 1.2 million shares.
Inventory was $1.84 billion compared to $1.92 billion in the previous quarter.
The company’s shares started trading on a 10-for-1 stock split basis on July 15, 2024 and also filed a mixed shelf offering on July 08.
Segments
Infrastructure Software
Infrastructure Software revenue grew by 175% YoY to $5.29 billion, primarily due to the contribution of VMware, accelerating from 153% growth in the previous quarter. Organically it grew by 35% YoY. The segment’s adjusted gross margins were 88% compared to 92% in the same period last year. The adjusted operating margin was 60% and excluding transition costs was 64% compared to 73% in the same period last year. The drop in margins was primarily due to VMware’s lower margin profile.
Management provided a key update on VMware in FQ2. “VMware revenue in Q1 was $2.1 billion, grew to $2.7 billion in Q2 and will accelerate towards a $4 billion per quarter run rate. We therefore expect operating margins for VMware to begin to converge towards that of classic Broadcom software by fiscal 2025.”
To illustrate VMware’s successful integration, management highlighted the streamlining of product SKUs from over 8,000 disparate SKUs to 4 core product offerings, thereby eliminating massive channel conflicts.
The company is also transitioning all VMware products to a subscription licensing model. “We are making good progress in transitioning all VMware products to a subscription licensing model. And since closing the deal, we have actually signed up close to 3,000 of our largest 10,000 customers to enable them to build a self-service virtual private cloud on-prem. Each of these customers typically sign up to a multiyear contract, which we normalize into an annual measure known as annualized booking value or ABV. This metric, ABV, for VMware products accelerated from $1.2 billion in Q1 to $1.9 billion in Q2. For a reference, for the consolidated Broadcom software portfolio, ABV grew from $1.9 billion in Q1 to $2.8 billion over the same period in Q2.”
Semiconductor Solutions
Semiconductor Solutions revenue grew by 6% YoY to $7.20 billion, accelerating from 4% in the previous quarter. The segment’s adjusted gross margins were 67%, down 370 basis points YoY, primarily due to a higher mix of custom AI accelerators. The adjusted operating margin was 55% compared to 59% in the same period last year. According to Oppenheimer, Semiconductor revenue is expected to grow 7% YoY and 2% sequentially in FQ3.
Networking revenue grew by 44% YoY to $3.8 billion, representing 53% of semiconductor revenue, led by strong demand from hyperscalers for AI networking and custom accelerators. According to Oppenheimer, Networking revenue is expected to grow 43% YoY and 5% QoQ in FQ3.
The company doubled the number of switches sold YoY, particularly the PAM-5 and Jericho3. It also benefits from the rapid transition of optical interconnects in AI data centers to 800 gigabit bandwidth.
CEO Hock Tan said in the earnings call. “Next year, we expect all mega-scale GPU deployments to be on Ethernet. We expect the strength in AI to continue, and because of that, we now expect networking revenue to grow 40% year-on-year compared to our prior guidance of over 35% growth.”
Server storage revenue declined by (-27%) YoY to $824 million. Management believes Q2 was the bottom in server storage and expects a modest recovery in the second half of the year. They expect server storage revenue to decline around 20% YoY range for FY2024 from the earlier mid-20 percentage range.
Broadband revenue declined by (-39%) YoY to $730 million due to the continued slowdown in telco spending. Management expects Broadband to bottom in the second half of the year with a recovery in 2025. They expect Broadband to decline to a high 30s percentage from the prior guide of just over 30% YoY decline.
Wireless revenue grew 2% YoY and down seasonally (-19%) sequentially to $1.6 billion. Management reiterated the previous guidance of flat YoY wireless revenue for FY2024.
Industrial resale revenue declined by (-10%) YoY to $234 million. They expect to be down double-digit YoY from the prior guide of high single-digit decline.
AI Revenue
AI revenue grew by 280% YoY and 35% sequentially to $3.1 billion. Management has increased the FY2024 revenue guide from over $10 billion to over $11 billion. J.P. Morgan analyst Harlan Sur is bullish on the AI opportunity and estimates about $12 billion in AI revenue this year and more than $16 billion next year. He also highlighted that Broadcom has recently won OpenAI’s first and second generation AI ASIC orders, making OpenAI the fourth major AI ASIC customer for the company.
The company highlighted their expertise in Ethernet and the opportunity, as they expect all mega GPU deployments to be on Ethernet. “Talking of AI accelerators, you may know our hyperscale customers are accelerating their investments to scale up the performance of these clusters. And to that end, we have just been awarded the next generation custom AI accelerators for these hyperscale customers of ours. Networking these AI accelerators is very challenging, but the technology does exist today. In Broadcom, with the deepest and broadest understanding of what it takes for complex, large workloads to be scaled out in an AI fabric. Proof in point, seven of the largest eight AI clusters in deployment today use Broadcom Ethernet solutions.”
Valuation
The company trades at a P/E ratio of 65.8 and a forward P/E ratio of 32.2, higher than the 5-year average of 41.1. Similarly, it trades at a P/S ratio of 16.1 and a forward P/S ratio of 13.8, higher than the average of 8.6. Valuation is a concern for all semiconductor stocks. At the same time, the market is rewarding the company with a premium valuation due to its transition from a value stock to an AI growth stock.
Conclusion
Strong AI growth, merger synergies, and the Ethernet opportunity make Broadcom a leading AI juggernaut second only to Nvidia. We will look toward the report to confirm our understanding that the fundamentals are on track due to ASICs growth, the acceleration in the VMWare software opportunity and networking.
Royston Roche, Equity Analyst at the I/O Fund, contributed to this article.
Broadcom will release its Q3 FY2024 results after market close today. Investors will be closely watching the AI revenue updates for the ASIC market leader. Broadcom has the potential to grab the attention after Nvidia due to the company’s unique position in the AI Infrastructure space. Recently Citi Analyst also highlighted that the company is catching up on Nvidia as the top holding as it adds more AI customers and accretion from VMware. In addition, it could benefit from investor fatigue with Nvidia, which aligns with our thoughts that we have highlighted here.
The main highlight in the last earnings call was the AI revenue FY2024 guide increase from over $10 billion to over $11 billion. The company beat the top-line and bottom-line estimates and also raised the full-year revenue guidance to $51 billion from the earlier $50 billion. Analysts believe the guide was conservative due to the company ramping up new ASIC customers, the potential bottoming of non-AI networking markets, and VMware integration progressing well. The company also raised the FY2024 adjusted EBITDA guide from 60% to 61%, which is positive as margin recovery post-VMware integration is also important for the stock momentum to grind higher.
Revenue
FQ2 revenue grew by 43% YoY to $12.49 billion, up from 34.2% growth in FQ1. Next quarter, revenue is going to accelerate to 46% YoY growth to $12.96 billion and further accelerate to 51% growth in FQ4.
FQ2 was the first full quarter with a contribution from VMware, and organically, it grew by 12%. Revenue beat estimates by 4.0% and was primarily helped by the strong 280% YoY growth in AI revenue of $3.1 billion, offset by cyclical weakness in enterprises and telcos. Management hinted that non-AI semiconductor revenue had bottomed out in Q2 and was likely to have a modest recovery in the second half of the year.
Management increased the FY2024 revenue guide from $50 billion to $51 billion Analysts expect FY2024 revenue to grow 43.8% YoY to $51.51 billion, organic growth is about 10%.
Analysts expect FY2025 revenue to grow 16.8% YoY to $60.15 billion and FY2026 revenue to grow 11.6% to $67.13 billion.
Ji Yoo, Head of Investor Relations, said in the earnings call, “For fiscal '24, we expect revenue from AI to be much stronger at over $11 billion. Non-AI semiconductor revenue has bottomed in Q2 and is likely to recover modestly for the second half of fiscal '24.
On infrastructure software, we're making very strong progress in integrating VMware and accelerating its growth. Pulling all these three key factors together, we are raising our fiscal '24 revenue guidance to $51 billion.”
Margins
The merger integration process will initially drag the margins in FY2024 due to transition costs and VMware’s lower margin profile. However, cost cutting and merger synergies are anticipated to improve margins in the long term.
The FQ2 gross margin was 62.3%, down from 70% in the same period last year and up from 61.7% in the previous quarter. Adjusted gross margin was 76.2%, down from 75.6% last year and up from 75.4% in the previous quarter.
Operating margin was 23.7%, down from 45.9% last year and up from 17.4% in the previous quarter. The operating margin was mainly lower from last year due to the increase in amortization of acquisition-related intangible assets, restructuring charges, and stock-based compensation.
The adjusted operating margin was 57.2%, down from 62% last year and up marginally from 57.1% in the previous quarter. Excluding transition costs, the adjusted operating margin was 59% and remained the same as in the previous quarter.
Net income was $2.1 billion or 17% of revenue compared to $3.48 billion or 39.9% of revenue in the same period last year. The lower net income was mainly due to the points discussed in the above paragraphs and higher interest expenses this year. The adjusted net income was $5.39 billion or 43.2% of revenue compared to $4.49 billion or 51.4% of revenue last year.
EPS
GAAP EPS was $0.44 compared to $0.82 in the same period last year. The adjusted EPS grew by 6.2% YoY to $1.096 and beat estimates by 1.1%, helped by cost savings. The company has been able to reduce VMware spending to $1.6 billion from the previous $2.3 billion pre-acquisition. Management expects to exit Q4 with a spending of $1.3 billion run rate, better than the previous plan of $1.4 billion. It is further expected to stabilize at $1.2 billion post-integration.
Analysts expect adjusted EPS to accelerate to 14.3% YoY growth to $1.20 in FQ3 and to 23.7% YoY growth to $1.37 in FQ4.
Analysts expect FY2024 adjusted EPS to grow 12.4% YoY to $4.75 and accelerate to 27.7% growth to $6.06 in FY2025.
FQ2 adjusted EBITDA was 59.5%, compared to 65.1% in the same period last year and 59.8% in the previous quarter. The drop is mainly due to VMware's lower margin. The post-integration is progressing well, and management also raised the FY2024 adjusted EBITDA guide from 60% to 61%. They also expect VMware's adjusted operating margin to match Broadcom’s software margin by FY2025.
Cash Flow and Balance Sheet
The company has high debt as it has been growing through successful acquisitions. While high debt is a concern, the company is focusing on repaying medium-term debt and has strong cash flows. Also, the company’s debt prior to the VMware acquisition has long maturities.
Operating cash flow was $4.58 billion or 36.7% of revenue compared to $4.5 billion or 51.6% of revenue in the same period last year and 40.3% in the previous quarter.
Free cash flow was $4.45 billion or 35.6% of revenue compared to $4.38 billion or 50.2% of revenue last year and 39.2% in the previous quarter. Free cash flow excluding cash used for restructuring and integration was $5.3 billion or 42% of revenue. Free cash flow as a percentage of revenue declined from last year due to higher interest expenses related to debt for VMware acquisition and “higher cash taxes due to a higher mix of US income and the delay in the reenactment of Section 174.”
Cash was $9.81 billion and debt of $74.02 billion compared to $11.9 billion and $75.9 billion in the previous quarter. The weighted average coupon rate and term to maturity of $48 billion fixed rate debt is 3.5% and 8.2 years, respectively. The weighted average coupon rate and term to maturity of floating rate debt are 6.6% and 2.8 years, respectively. The company repaid $2 billion of floating rate debt in FQ2 and plans to maintain this quarterly repayment throughout FY2024.
The company paid $2.4 billion in dividends and $1.5 billion in withholding taxes due to the vesting of employee equity, eliminating 1.2 million shares.
Inventory was $1.84 billion compared to $1.92 billion in the previous quarter.
The company’s shares started trading on a 10-for-1 stock split basis on July 15, 2024 and also filed a mixed shelf offering on July 08.
Segments
Infrastructure Software
Infrastructure Software revenue grew by 175% YoY to $5.29 billion, primarily due to the contribution of VMware, accelerating from 153% growth in the previous quarter. Organically it grew by 35% YoY. The segment’s adjusted gross margins were 88% compared to 92% in the same period last year. The adjusted operating margin was 60% and excluding transition costs was 64% compared to 73% in the same period last year. The drop in margins was primarily due to VMware’s lower margin profile.
Management provided a key update on VMware in FQ2. “VMware revenue in Q1 was $2.1 billion, grew to $2.7 billion in Q2 and will accelerate towards a $4 billion per quarter run rate. We therefore expect operating margins for VMware to begin to converge towards that of classic Broadcom software by fiscal 2025.”
To illustrate VMware’s successful integration, management highlighted the streamlining of product SKUs from over 8,000 disparate SKUs to 4 core product offerings, thereby eliminating massive channel conflicts.
The company is also transitioning all VMware products to a subscription licensing model. “We are making good progress in transitioning all VMware products to a subscription licensing model. And since closing the deal, we have actually signed up close to 3,000 of our largest 10,000 customers to enable them to build a self-service virtual private cloud on-prem. Each of these customers typically sign up to a multiyear contract, which we normalize into an annual measure known as annualized booking value or ABV. This metric, ABV, for VMware products accelerated from $1.2 billion in Q1 to $1.9 billion in Q2. For a reference, for the consolidated Broadcom software portfolio, ABV grew from $1.9 billion in Q1 to $2.8 billion over the same period in Q2.”
Semiconductor Solutions
Semiconductor Solutions revenue grew by 6% YoY to $7.20 billion, accelerating from 4% in the previous quarter. The segment’s adjusted gross margins were 67%, down 370 basis points YoY, primarily due to a higher mix of custom AI accelerators. The adjusted operating margin was 55% compared to 59% in the same period last year. According to Oppenheimer, Semiconductor revenue is expected to grow 7% YoY and 2% sequentially in FQ3.
Networking revenue grew by 44% YoY to $3.8 billion, representing 53% of semiconductor revenue, led by strong demand from hyperscalers for AI networking and custom accelerators. According to Oppenheimer, Networking revenue is expected to grow 43% YoY and 5% QoQ in FQ3.
The company doubled the number of switches sold YoY, particularly the PAM-5 and Jericho3. It also benefits from the rapid transition of optical interconnects in AI data centers to 800 gigabit bandwidth.
CEO Hock Tan said in the earnings call. “Next year, we expect all mega-scale GPU deployments to be on Ethernet. We expect the strength in AI to continue, and because of that, we now expect networking revenue to grow 40% year-on-year compared to our prior guidance of over 35% growth.”
Server storage revenue declined by (-27%) YoY to $824 million. Management believes Q2 was the bottom in server storage and expects a modest recovery in the second half of the year. They expect server storage revenue to decline around 20% YoY range for FY2024 from the earlier mid-20 percentage range.
Broadband revenue declined by (-39%) YoY to $730 million due to the continued slowdown in telco spending. Management expects Broadband to bottom in the second half of the year with a recovery in 2025. They expect Broadband to decline to a high 30s percentage from the prior guide of just over 30% YoY decline.
Wireless revenue grew 2% YoY and down seasonally (-19%) sequentially to $1.6 billion. Management reiterated the previous guidance of flat YoY wireless revenue for FY2024.
Industrial resale revenue declined by (-10%) YoY to $234 million. They expect to be down double-digit YoY from the prior guide of high single-digit decline.
AI Revenue
AI revenue grew by 280% YoY and 35% sequentially to $3.1 billion. Management has increased the FY2024 revenue guide from over $10 billion to over $11 billion. J.P. Morgan analyst Harlan Sur is bullish on the AI opportunity and estimates about $12 billion in AI revenue this year and more than $16 billion next year. He also highlighted that Broadcom has recently won OpenAI’s first and second generation AI ASIC orders, making OpenAI the fourth major AI ASIC customer for the company.
The company highlighted their expertise in Ethernet and the opportunity, as they expect all mega GPU deployments to be on Ethernet. “Talking of AI accelerators, you may know our hyperscale customers are accelerating their investments to scale up the performance of these clusters. And to that end, we have just been awarded the next generation custom AI accelerators for these hyperscale customers of ours. Networking these AI accelerators is very challenging, but the technology does exist today. In Broadcom, with the deepest and broadest understanding of what it takes for complex, large workloads to be scaled out in an AI fabric. Proof in point, seven of the largest eight AI clusters in deployment today use Broadcom Ethernet solutions.”
Valuation
The company trades at a P/E ratio of 65.8 and a forward P/E ratio of 32.2, higher than the 5-year average of 41.1. Similarly, it trades at a P/S ratio of 16.1 and a forward P/S ratio of 13.8, higher than the average of 8.6. Valuation is a concern for all semiconductor stocks. At the same time, the market is rewarding the company with a premium valuation due to its transition from a value stock to an AI growth stock.
Conclusion
Strong AI growth, merger synergies, and the Ethernet opportunity make Broadcom a leading AI juggernaut second only to Nvidia. We will look toward the report to confirm our understanding that the fundamentals are on track due to ASICs growth, the acceleration in the VMWare software opportunity and networking.
Royston Roche, Equity Analyst at the I/O Fund, contributed to this article.
This article was originally published on Forbes on Updated Aug 30, 2024, 08:59am EDTForbesForbes on Updated Aug 30, 2024, 08:59am EDT
Our firm extrapolated supply chain data to conclude that Blackwell is in production at TSM and SMCI last week in the analysis: Nvidia Stock: Blackwell Suppliers Shrug Off Delay. The media was making much ado about nothing (and astonishingly, still is) despite crystal clear confirmation from Nvidia’s management team that all is well.
Given these delay rumors, it was widely expected that Nvidia’s management would provide some transparency in Q2 as to the status of Blackwell. I joined ‘Making Money’ on Fox Business Network shortly before Nvidia’s report, telling host Charles Payne that “we are getting bullish signals from the supply chain,” such as TSM’s HPC growth and Super Micro’s liquid cooling growth, and that I “fully expect Nvidia’s management team to calm any concerns about the outlook for Blackwell.”
Direct liquid cooling doesn’t lie as it’s intricately linked to the Blackwell launch, implying that Blackwell would indeed ship by Q4 – and Nvidia just confirmed that (multiple times) in Q2’s release:
“Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026. In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”
Later in the call, Jensen Huang stated: “There were no functional changes necessary. And so we're sampling functional samples of Blackwell — Grace Blackwell in a variety of system configurations as we speak. There are something like 100 different types of Blackwell-based systems that are built that were shown at Computex. And we're enabling our ecosystem to start sampling those. The functionality of Blackwell is as it is, and we expect to start production in Q4.”
We had published for our free readers going into the print that the valuation was stretched, and it would require fiscal year revisions to create room in the valuation. As you’ll see below, we got a few revisions today, which is paramount for the stock price. Will these upward revisions be enough to sustain the price? We look at this and more below.
Q2 Revenue Beats Estimates
Q2’s revenue of $30.04 billion increased 122% YoY and 15% QoQ, with management pointing out that “customers continue to accelerate their Hopper architecture purchases while gearing up to adopt Blackwell.” This marked a $1.3 billion beat to the consensus estimate for $28.75 billion. It also was a deceleration from 262% YoY growth in Q1, as Nvidia is now facing tougher comps against the vertical ramp of Hopper last year. GAAP EPS of $0.67 beat estimates by $0.03, and represented YoY growth of 168% and QoQ growth of 12%.
Nvidia guided for Q3 revenue of $32.5 billion, once again above consensus estimates, though it was only $700 million higher than the $31.77 billion estimate at the midpoint. This represents growth of 79.4% YoY at midpoint, compared to the estimate for 75.3% growth next quarter. Despite this being one of the ‘smaller’ beats in recent quarters, it’s a testament to the strength of Nvidia’s demand to guide for $2.5 billion sequential growth primarily based on Hopper demand with no contribution from Blackwell.
Source: I/O Fund
Sign up for I/O Fund's free newsletter with gains of up to 2600% because of Nvidia's epic run – Click hereClick hereClick here
Data Center Strength Visible with Blackwell on Tap
Data center revenue surpassed a $105 billion annualized run rate this quarter, up from $90 billion annualized last quarter, as Nvidia reported $26.27 billion in data center revenue, up 152% YoY and 16% QoQ. Nvidia said that “Hopper demand is strong, and shipments are expected to increase in the second half of fiscal 2025,” while Blackwell is on track to ramp in Q4 with “several” billions in revenue expected that quarter.
Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.
In the segment, compute revenue was $22.6 billion, up 162% YoY, while networking revenue was $3.67 billion, up 114% YoY. In networking, Nvidia noted that InfiniBand and Ethernet drove growth in the quarter, and the 16% QoQ growth included “a doubling of Ethernet for AI revenue.”
Source: I/O Fund
Nvidia’s Q3 revenue guide implies data center revenue above $28 billion to $28.5 billion, which we had modeled in our pre-earnings analysis earlier this week.
Delay Concerns Cleared, But Valuation Looks Stretched
Nvidia cleared the delay concerns for Blackwell, saying that they “shipped customer samples of our Blackwell architecture in the second quarter. We executed a change to the Blackwell GPU mask to improve production yield. Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026,” with several billion in Blackwell revenue expected in Q4. Purchase commitments reiterated that Nvidia is serious about launching on schedule, and lining up the capacity and components to launch in full-force by the end of the year.
I spoke with Yahoo Finance on Thursday morning following the report, reemphasizing that the delay concerns were “completely thrown off the table last night. … Wall Street obviously is very closely tied to estimates, and we never saw revisions downward based on the so-called delay. … Nvidia beat, and they’re saying Blackwell is basically on time,” which is “not a concern — if anything, it’s extremely bullish.”
"Blackwell is not a concern of anything," I/O Fund lead tech analyst @Beth_Kindig says following $NVDA earnings. "It's extremely bullish."
However, I cautioned on the valuation: “When you have a high-flyer like Nvidia, you get stretched at times. Going into the print, we warned our members that this valuation is looking a little toppy. What we need is for the fiscal year estimates next year to go up, so we’re in a waiting game for analysts to revise their estimates upward, which eventually they will, but until then the valuation is stretched.”
This morning, while I was being interviewed by Yahoo, we’ve already seen analyst estimates for Nvidia’s revenue revised higher following the report:
Fiscal 2025 revenue is now estimated at $124.8 billion, up 3.9% from the $120.1 billion estimate prior to Q2’s report.
Fiscal 2026 revenue is now estimated at $172.1 billion, up 5.2% from the $163.6 billion estimate prior to the report.
However, the true impact of Blackwell is yet to be seen in these estimates, with the only clues right now being Q3’s $32.5 billion guide and expectations for several billion in Blackwell revenue in Q4. From a long-term perspective, I explained on Yahoo Finance that the first “pathway for growth is to pay very close attention to Nvidia around the fiscal year guide,” while the “second-biggest moment of the year will be when Blackwell is shipping in volume. This will be the Q2 report, but we’ll get some signs in Q1 with that forward guide.” I believe that “early next year will be fireworks” for Nvidia, similar to Hopper’s moment in the fiscal Q1 report in May 2023.
Every Thursday at 4:30 pm Eastern, the I/O Fund team holds a webinar for premium members to discuss how to navigate the broad market, as well as various stock entries and exits. We offer trade alerts plus an automated hedging signal. The I/O Fund team is one of the only audited portfolios available to individual investors. Learn more here.Learn more hereLearn more here.
Eyes on Margins as Blackwell Ramps
Margins remained strong in Q2, with Nvidia reporting gross and operating margins at the high end and above its guided ranges. However, management guided for Q3 margins to contract slightly QoQ, suggesting that Q1 was the peak for both gross and operating margins with some pressure ahead as Blackwell gears up to launch in Q4.
GAAP gross margin was 75.1% in Q2, ahead of management’s guide for 74.8%. Adjusted gross margin was 75.7%, ahead of guidance for 75.5%. Per the CFO: “As our Data Center mix continues to shift to new products, we expect this trend to continue into the fourth quarter of fiscal 2025.” It’s likely she is referring to the higher cost of memory components, which we outlined in our pre-earnings analysis.
GAAP operating margin was 62.1%, ahead of the implied guide for 60.5%, indicative of the operating leverage power that Nvidia still commands in mid-launch cycle for Hopper with the H200s shipping now. Adjusted operating margin was 66.4%, ahead of the implied guide of 65.5%.
GAAP net margin was 55.3% down from 57.1% last quarter. This represents profits of $16.6 billion, up over $2 billion. This was a very large beat compared to the $14.3 billion guided.
Source: I/O Fund
The chart above shows Nvidia’s margins, with the slight sequential contraction this quarter and next quarter visible. It’s no small feat to maintain GAAP operating margin >60% for four consecutive quarters while simultaneously undergoing the semiconductor industry’s most advanced and most rapid product release cycle. However, with management guiding for full-year gross margins to be in the mid-70% range, we’ll be keeping a close eye on how margins trend in Q3 heading into Q4 as Blackwell ramps — where the market is a tad concerned is gross margins, which peaked at 78.4% and will exit the year in the mid-70% range.
Conclusion
Our pre-earnings writeup expressed concerns about the valuation going into the print, and I think the selling on Thursday reflects the valuation. Our firm stuck our neck out over the past few weeks to bring quality information to our readers on how the supply chain for Blackwell is ramping. We were the first and only firm that I’m aware of to present actionable data that countered what other media outlets were reporting. To refresh your memory, media outlets stated Blackwell was delayed into Q1: “If the upcoming AI chips, known as the B100, B200 and GB200, are delayed three months or more, it may prevent some customers from operating large clusters of the chips in their data centers in the first quarter of 2025, as they had planned.”
In contrast, my analysis stated: “From the horse’s mouth, Nvidia’s own management team, it was stated during the GTC Financial Analyst Day in March that the very first systems will ship in Q4, but to expect constraints.”
Well, we have our answer – Blackwell is, in fact, shipping in Q4 and ramping in Q1. Purchase commitments up 48% QoQ help to reflect how serious the company is when it comes to the speed of ramping shipments.
Direct liquid cooling doesn't lie.
My firm pieced together that Blackwell was, in fact, shipping in Q4 — proving my firm’s never-ending tenacity on this stock.
Earnings reports are truly 50/50 – nobody can tell you what the market will do following a report. For example, we had high confidence Nvidia would beat, but there’s much more to consider than a beat. What’s important is to have a strategy. Our firm champions actively managing tech positions rather than buy-and-hold. Our plan is to trim Nvidia at key levels and attempt to buy lower. This is due to valuation concerns, but also importantly, many AI stocks are trading at stretched valuations. We’ve stated publicly a few times that Nvidia is a buy on dips, implying investors who are patient will find entries at lower prices.
Please note: The I/O Fund conducts research and draws conclusions for the company’s portfolio. We then share that information with our readers and offer real-time trade notifications. This is not a guarantee of a stock’s performance and it is not financial advice. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis. Beth Kindig and the I/O Fund own shares in NVDA at the time of writing and may own stocks pictured in the charts.
Given the rumors that had transpired about Blackwell’s delays, it was widely expected that Nvidia’s management would provide some transparency in Q2 as to the status of Blackwell. We had tracked a few key data points in the supply chain, such as TSM’s HPC growth and Super Micro’s liquid cooling growth. Direct liquid cooling doesn’t lie as it’s intricately linked to the Blackwell launch, implying that Blackwell would indeed ship by Q4 – and Nvidia just confirmed that:
“Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026. In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”
Nvidia also guided for Q3 revenue of $32.5 billion, once again above consensus estimates, though it was only $700 million higher at the midpoint; despite this being one of the ‘smaller’ beats, it’s a testament to the strength of Nvidia’s demand to guide for $2.5 billion sequential growth primarily based on Hopper demand with no contribution from Blackwell.
Revenue and EPS
Revenue of $30.04 billion increased 122% YoY and 15% QoQ, with management pointing out that “customers continue to accelerate their Hopper architecture purchases while gearing up to adopt Blackwell.” This marked a $1.3 billion beat to the consensus estimate for $28.75 billion. It also was a deceleration from 262% YoY growth in Q1, as Nvidia is now facing tougher comps against the vertical ramp of Hopper last year.
For Q3, Nvidia guided for revenue of $32.5 billion, +/- 2%, ahead of the consensus estimate for $31.77 billion. This represents growth of 79.4% YoY at midpoint, compared to the estimate for 75.3% growth next quarter.
GAAP EPS of $0.67 beat estimates by $0.06, and represented YoY growth of 168% and QoQ growth of 12%.
Adjusted EPS of $0.68 beat estimates by $0.04, and represented YoY growth of 152% and QoQ growth of 11%.
Key Segments
Data center revenue surpassed a $100 billion annualized run rate this quarter, up from $90 billion annualized, as Nvidia reported $26.27 billion in data center revenue, up 152% YoY and 16% QoQ. Nvidia said that “Hopper demand is strong, and shipments are expected to increase in the second half of fiscal 2025,” while Blackwell is on track to ramp in Q4 with “several” billions in revenue expected that quarter.
Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.
In the segment, compute revenue was $22.6 billion, up 162% YoY, while networking revenue was $3.67 billion, up 114% YoY. In networking, Nvidia noted that InfiniBand and Ethernet drove growth in the quarter, and the 16% QoQ growth included “a doubling of Ethernet for AI revenue.”
Nvidia’s Q3 revenue guide implies data center revenue slightly above $28 billion to $28.5 billion, which we had modeled in our pre-earnings analysis earlier this week.
Gaming revenue increased 16% YoY and 9% QoQ to $2.88 billion, driven by increased sales of GeForce RTX 40 Series chips.
Automotive revenue increased 37% YoY and 5% QoQ to $346 million, driven by AI cockpit and self-driving solutions.
Pro-viz revenue increased 20% YoY and 6% QoQ to $454 million.
OEM and other revenue increased 33% YoY and 13% QoQ to $88 million.
Margins
Margins remained strong in Q2, with Nvidia reporting gross and operating margins at the high end and above its guided ranges. However, management guided for Q3 margins to contract slightly QoQ, suggesting that Q1 was the peak for both gross and operating margins with some pressure ahead as Blackwell gears up to launch in Q4.
GAAP gross margin was 75.1% in Q2, ahead of management’s guide for 74.8%. Adjusted gross margin was 75.7%, ahead of guidance for 75.5%.
Per the CFO: “As our Data Center mix continues to shift to new products, we expect this trend to continue into the fourth quarter of fiscal 2025.” It’s likely she is referring to the higher cost of memory components, which we outlined in the pre-earnings analysis.
GAAP operating margin was 62.1%, ahead of the implied guide for 60.5%, indicative of the operating leverage power that Nvidia still commands in mid-launch cycle for Hopper with the H200s shipping now. Adjusted operating margin was 66.4%, ahead of the implied guide of 65.5%.
Per the CFO: “Sequentially, GAAP and non-GAAP operating expenses were up 12%, primarily reflecting higher compensation-related costs.”
GAAP net margin was 55.3% down from 57.1% last quarter. This represents profits of $16.6 billion, up over $2 billion. This was a very large beat compared to the $14.3 billion guided.
The chart below shows Nvidia’s margins, with the slight sequential contraction this quarter and next quarter visible. It’s no small feat to maintain GAAP operating margin >60% for four consecutive quarters while simultaneously undergoing the semiconductor industry’s most advanced and most rapid product release cycle. However, with management guiding for full-year gross margins to be in the mid-70% range, we’ll be keeping a close eye on how margins trend in Q3 heading into Q4 as Blackwell ramps.
Cash Flows
For the first time since Hopper stole the show in early 2023 with red-hot growth, Nvidia’s cash flows declined sequentially, with cash flow margins also shrinking. Nvidia said the sequential declines were caused by higher cash taxes paid.
Operating cash flow was $14.49 billion, up more than 128% YoY but down (5.6%) QoQ. This was the first quarter with a sequential decline in OCF since Q3 FY23.
Operating cash flow margin was 48.2%, down from 58.9% last quarter but up from 47.0% in the year ago quarter.
Free cash flow was $13.48 billion, up 123% YoY but down nearly (10%) QoQ.
Free cash flow margin was 44.9%, down from 57.3% last quarter and nearly flat with 44.8% in the year ago quarter.
Nvidia had $34.8 billion in cash, equivalents and marketable securities on hand, and debt totaled $8.46 billion.
Nvidia used cash of $7.4 billion toward shareholder returns, with $246 million in dividends paid and $7.2 billion in shares repurchased in Q2. Nvidia’s board also approved an additional $50 billion to its share repurchase authorization.
Earnings Call
Blackwell Shipping on Schedule
Our firm extrapolated supply chain data to conclude that Blackwell is in production at TSM and SMCI last week in the analysis: Nvidia Stock: Blackwell Suppliers Shrug Off Delay. This was not an easy analysis to write as it meant going up against The Information, a highly regarded publication. Yet, I felt confident that I know Nvidia stock better than a reporter, so hey — why not throw my hat in the ring. From my vantage point, since I cover the universe so closely, it was becoming clear the “breaking news” was an exaggerated claim that was inaccurate and ultimately hurt a lot of investors.
The CFO printed commentary stated: “We shipped customer samples of our Blackwell architecture in the second quarter. We executed a change to the Blackwell GPU mask to improve production yield. Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026. In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”
During a question on Hopper, Jensen Huang confirmed Q1 will be when Blackwell is operable in data centers: “And although Blackwell will start shipping out in billions of dollars at the end of this year, the standing up of the capacity is still probably weeks and a month or so away.”
The most direct Q&A snippet regarding Blackwell’s timing was the following:
Question Vivek Arya (Analysts)
Jensen, you mentioned in the prepared comments that there's a change in the Blackwell GPU mask. I'm curious, are there any other incremental changes in back-end packaging or anything else? And I think related, you suggested that you could ship several billion dollars of Blackwell in Q4 despite the change in the design. Is it because all these issues will be solved by then? Just help us size what is the overall impact of any changes in Blackwell timing, what that means to your kind of revenue profile and how are customers reacting to it.
Answer Jensen Huang (Executives)
Yes. Thanks, Vivek. The change to the mask is complete. There were no functional changes necessary. And so we're sampling functional samples of Blackwell, Grace Blackwell, and a variety of system configurations as we speak. There are something like 100 different types of Blackwell-based systems that are built that were shown at Computex, and we're enabling our ecosystem to start sampling those. The functionality of Blackwell is as it is, and we expect to start production in Q4.
-End Quote
This was also a nice statement from Huang that we should memorialize in this earnings writeup – perhaps unrelated to the delay, but certainly a statement to capture for our Members: “Yes, next year is going to be a great year. We expect to grow our Data Center business quite significantly next year. Blackwell is going to be a complete game changer for the industry. And Blackwell is going to carry into the following year”
Direct Liquid Cooling (DLC)
DLC stocks have been hit hard lately, primarily Dell and most certainly Super Micro following the short report and the delayed filing of the 10-K. However, SMCI’s earnings report was quite clear DLC is ramping which is why we’ve persevered in our positions. An analyst asked about DLC and the potential for supply chain issues with the thermal management side. Jensen Huang’s answer in agreement with what we saw from SMCI’s report.
Note: the analyst in the Q&A points out the purchase commitments and supply obligations being bullish, we reported on this under Key Segments – this is an important takeaway from this earnings report that shows Blackwell is ramping nicely. This is what he is referring to — “Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.”Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.”
Question Timothy Arcuri (Analysts)
I had a question on the shape of the revenue growth, both near and longer term. I know Colette, you did increase OpEx for the year. And if I look at the increase in your purchase commitments and your supply obligations, that's also quite bullish.
On the other hand, there are some school who've thought that not that many customers really seem ready for liquid cooling, and I do recognize that some of these racks can be air cooled. But Jensen, is that something to consider sort of on the shape of how Blackwell is going to ramp? And then I guess when you look beyond next year, which is obviously going to be a great year and you look into '26, do you worry about any other gating factors like, say, the power supply chain or, at some point, models start to get smaller? I'm just wondering if you can speak to that.
Answer Jensen Huang (Executives)
[…] The Grace Blackwell is liquid cooled. However, the number of data centers that want to go to liquid cooled is quite significant. And the reason for that is because we can, in a liquid-cooled data center, in any power-limited data center, whatever size of data center you choose, you could install and deploy anywhere from 3 to 5x the AI throughput compared to the past. And so liquid cooling is cheaper. Our TCO is better, and liquid cooling allows you to have the benefit of this capability we call NVLink, which allows us to expand it to 72 Grace Blackwell packages, which has essentially 144 GPUs.
And so imagine 144 GPUs connected in NVLink. And we're increasingly showing you the benefits of that. And the next click is obviously very low latency, very high throughput large language model inference, and the large NVLink domain is going to be a game changer for that. And so I think people are very comfortable deploying both. And so almost every CSP we're working with are deploying some of both. And so I'm pretty confident that we'll ramp it up just fine.”
-End Quote
My note: having a hard time closing our weaker DLC positions based on what we know is coming. My main regret is not keeping some dry powder to buy at these lower levels.
Margins:
Discussion on margins felt a bit forced as Nvidia’s margins continue to beat the rest of the Mag 7. Consider that Nvidia’s GAAP operating margin this quarter was 62.1%:
Where the market is a tad concerned is gross margins, which peaked at 78.4% and will exit the year in the mid-70% range. We prepped our Members for this in the pre-earnings writeup. The CFO stated: “Your second piece is in terms of our gross margin. We provided gross margin for our Q3. We provided our gross margin on a non-GAAP at about 75%. We'll work with all the different transitions that we're going through, but we do believe we can do that 75% in Q3. We provided that we're still on track for the full year also in the mid-70s or approximately the 75%. So we're going to see some slight difference possibly in Q4, again with our transitions and the different cost structures that we have on our new product introductions.”
Conclusion:
Our pre-earnings writeup expressed concerns about the valuation going into the print. I think the selling after hours reflects the valuation.
Earnings reports are truly 50/50 – nobody can tell you what the market will do following a report. For example, we had high confidence Nvidia would beat, but there’s much more to consider than a beat.
What’s important is to have a strategy, and we do. Our strategy is to trim this position depending on how the price reacts at key levels. Our goal is to then re-allocate what we trim at lower levels. You will get trade alerts to this effect when we think the timing is right.
We recently discussed an emerging ad-tech AI leader in our Advanced Market Signals Tier. Our proprietary custom-built screeners identified the stock. The company witnessed a strong acceleration in growth and margins due to its AI-powered advertising engine. To read the article, upgrade here.
Damien Robbins, Equity Analyst for the I/O Fund, contributed to this analysis.
Given the rumors that had transpired about Blackwell’s delays, it was widely expected that Nvidia’s management would provide some transparency in Q2 as to the status of Blackwell. We had tracked a few key data points in the supply chain, such as TSM’s HPC growth and Super Micro’s liquid cooling growth. Direct liquid cooling doesn’t lie as it’s intricately linked to the Blackwell launch, implying that Blackwell would indeed ship by Q4 – and Nvidia just confirmed that:
“Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026. In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”
Nvidia also guided for Q3 revenue of $32.5 billion, once again above consensus estimates, though it was only $700 million higher at the midpoint; despite this being one of the ‘smaller’ beats, it’s a testament to the strength of Nvidia’s demand to guide for $2.5 billion sequential growth primarily based on Hopper demand with no contribution from Blackwell.
Revenue and EPS
Revenue of $30.04 billion increased 122% YoY and 15% QoQ, with management pointing out that “customers continue to accelerate their Hopper architecture purchases while gearing up to adopt Blackwell.” This marked a $1.3 billion beat to the consensus estimate for $28.75 billion. It also was a deceleration from 262% YoY growth in Q1, as Nvidia is now facing tougher comps against the vertical ramp of Hopper last year.
For Q3, Nvidia guided for revenue of $32.5 billion, +/- 2%, ahead of the consensus estimate for $31.77 billion. This represents growth of 79.4% YoY at midpoint, compared to the estimate for 75.3% growth next quarter.
GAAP EPS of $0.67 beat estimates by $0.03, and represented YoY growth of 168% and QoQ growth of 12%.
Adjusted EPS of $0.68 beat estimates by $0.04, and represented YoY growth of 152% and QoQ growth of 11%.
Key Segments
Data center revenue surpassed a $100 billion annualized run rate this quarter, up from $90 billion annualized, as Nvidia reported $26.27 billion in data center revenue, up 152% YoY and 16% QoQ. Nvidia said that “Hopper demand is strong, and shipments are expected to increase in the second half of fiscal 2025,” while Blackwell is on track to ramp in Q4 with “several” billions in revenue expected that quarter.
Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.
In the segment, compute revenue was $22.6 billion, up 162% YoY, while networking revenue was $3.67 billion, up 114% YoY. In networking, Nvidia noted that InfiniBand and Ethernet drove growth in the quarter, and the 16% QoQ growth included “a doubling of Ethernet for AI revenue.”
Nvidia’s Q3 revenue guide implies data center revenue slightly above $28 billion to $28.5 billion, which we had modeled in our pre-earnings analysis earlier this week.
Gaming revenue increased 16% YoY and 9% QoQ to $2.88 billion, driven by increased sales of GeForce RTX 40 Series chips.
Automotive revenue increased 37% YoY and 5% QoQ to $346 million, driven by AI cockpit and self-driving solutions.
Pro-viz revenue increased 20% YoY and 6% QoQ to $454 million.
OEM and other revenue increased 33% YoY and 13% QoQ to $88 million.
Margins
Margins remained strong in Q2, with Nvidia reporting gross and operating margins at the high end and above its guided ranges. However, management guided for Q3 margins to contract slightly QoQ, suggesting that Q1 was the peak for both gross and operating margins with some pressure ahead as Blackwell gears up to launch in Q4.
GAAP gross margin was 75.1% in Q2, ahead of management’s guide for 74.8%. Adjusted gross margin was 75.7%, ahead of guidance for 75.5%.
Per the CFO: “As our Data Center mix continues to shift to new products, we expect this trend to continue into the fourth quarter of fiscal 2025.” It’s likely she is referring to the higher cost of memory components, which we outlined in the pre-earnings analysis.
GAAP operating margin was 62.1%, ahead of the implied guide for 60.5%, indicative of the operating leverage power that Nvidia still commands in mid-launch cycle for Hopper with the H200s shipping now. Adjusted operating margin was 66.4%, ahead of the implied guide of 65.5%.
Per the CFO: “Sequentially, GAAP and non-GAAP operating expenses were up 12%, primarily reflecting higher compensation-related costs.”
GAAP net margin was 55.3% down from 57.1% last quarter. This represents profits of $16.6 billion, up over $2 billion. This was a very large beat compared to the $14.3 billion guided.
The chart below shows Nvidia’s margins, with the slight sequential contraction this quarter and next quarter visible. It’s no small feat to maintain GAAP operating margin >60% for four consecutive quarters while simultaneously undergoing the semiconductor industry’s most advanced and most rapid product release cycle. However, with management guiding for full-year gross margins to be in the mid-70% range, we’ll be keeping a close eye on how margins trend in Q3 heading into Q4 as Blackwell ramps.
Cash Flows
For the first time since Hopper stole the show in early 2023 with red-hot growth, Nvidia’s cash flows declined sequentially, with cash flow margins also shrinking. Nvidia said the sequential declines were caused by higher cash taxes paid.
Operating cash flow was $14.49 billion, up more than 128% YoY but down (5.6%) QoQ. This was the first quarter with a sequential decline in OCF since Q3 FY23.
Operating cash flow margin was 48.2%, down from 58.9% last quarter but up from 47.0% in the year ago quarter.
Free cash flow was $13.48 billion, up 123% YoY but down nearly (10%) QoQ.
Free cash flow margin was 44.9%, down from 57.3% last quarter and nearly flat with 44.8% in the year ago quarter.
Nvidia had $34.8 billion in cash, equivalents and marketable securities on hand, and debt totaled $8.46 billion.
Nvidia used cash of $7.4 billion toward shareholder returns, with $246 million in dividends paid and $7.2 billion in shares repurchased in Q2. Nvidia’s board also approved an additional $50 billion to its share repurchase authorization.
Earnings Call
Blackwell Shipping on Schedule
Our firm extrapolated supply chain data to conclude that Blackwell is in production at TSM and SMCI last week in the analysis: Nvidia Stock: Blackwell Suppliers Shrug Off Delay. This was not an easy analysis to write as it meant going up against The Information, a highly regarded publication. Yet, I felt confident that I know Nvidia stock better than a reporter, so hey — why not throw my hat in the ring. From my vantage point, since I cover the universe so closely, it was becoming clear the “breaking news” was an exaggerated claim that was inaccurate and ultimately hurt a lot of investors.
The CFO printed commentary stated: “We shipped customer samples of our Blackwell architecture in the second quarter. We executed a change to the Blackwell GPU mask to improve production yield. Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal 2026. In the fourth quarter, we expect to ship several billion dollars in Blackwell revenue.”
During a question on Hopper, Jensen Huang confirmed Q1 will be when Blackwell is operable in data centers: “And although Blackwell will start shipping out in billions of dollars at the end of this year, the standing up of the capacity is still probably weeks and a month or so away.”
The most direct Q&A snippet regarding Blackwell’s timing was the following:
Question Vivek Arya (Analysts)
Jensen, you mentioned in the prepared comments that there's a change in the Blackwell GPU mask. I'm curious, are there any other incremental changes in back-end packaging or anything else? And I think related, you suggested that you could ship several billion dollars of Blackwell in Q4 despite the change in the design. Is it because all these issues will be solved by then? Just help us size what is the overall impact of any changes in Blackwell timing, what that means to your kind of revenue profile and how are customers reacting to it.
Answer Jensen Huang (Executives)
Yes. Thanks, Vivek. The change to the mask is complete. There were no functional changes necessary. And so we're sampling functional samples of Blackwell, Grace Blackwell, and a variety of system configurations as we speak. There are something like 100 different types of Blackwell-based systems that are built that were shown at Computex, and we're enabling our ecosystem to start sampling those. The functionality of Blackwell is as it is, and we expect to start production in Q4.
-End Quote
This was also a nice statement from Huang that we should memorialize in this earnings writeup – perhaps unrelated to the delay, but certainly a statement to capture for our Members: “Yes, next year is going to be a great year. We expect to grow our Data Center business quite significantly next year. Blackwell is going to be a complete game changer for the industry. And Blackwell is going to carry into the following year”
Direct Liquid Cooling (DLC)
DLC stocks have been hit hard lately, primarily Dell and most certainly Super Micro following the short report and the delayed filing of the 10-K. However, SMCI’s earnings report was quite clear DLC is ramping which is why we’ve persevered in our positions. An analyst asked about DLC and the potential for supply chain issues with the thermal management side. Jensen Huang’s answer in agreement with what we saw from SMCI’s report.
Note: the analyst in the Q&A points out the purchase commitments and supply obligations being bullish, we reported on this under Key Segments – this is an important takeaway from this earnings report that shows Blackwell is ramping nicely. This is what he is referring to — “Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.”Notably, purchase commitments and obligations for inventory and capacity rose nearly 48% QoQ to $27.8 billion, including “new commitments for Blackwell capacity and components,” another signal that Nvidia is prepared to ramp in full-force come Q4.”
Question Timothy Arcuri (Analysts)
I had a question on the shape of the revenue growth, both near and longer term. I know Colette, you did increase OpEx for the year. And if I look at the increase in your purchase commitments and your supply obligations, that's also quite bullish.
On the other hand, there are some school who've thought that not that many customers really seem ready for liquid cooling, and I do recognize that some of these racks can be air cooled. But Jensen, is that something to consider sort of on the shape of how Blackwell is going to ramp? And then I guess when you look beyond next year, which is obviously going to be a great year and you look into '26, do you worry about any other gating factors like, say, the power supply chain or, at some point, models start to get smaller? I'm just wondering if you can speak to that.
Answer Jensen Huang (Executives)
[…] The Grace Blackwell is liquid cooled. However, the number of data centers that want to go to liquid cooled is quite significant. And the reason for that is because we can, in a liquid-cooled data center, in any power-limited data center, whatever size of data center you choose, you could install and deploy anywhere from 3 to 5x the AI throughput compared to the past. And so liquid cooling is cheaper. Our TCO is better, and liquid cooling allows you to have the benefit of this capability we call NVLink, which allows us to expand it to 72 Grace Blackwell packages, which has essentially 144 GPUs.
And so imagine 144 GPUs connected in NVLink. And we're increasingly showing you the benefits of that. And the next click is obviously very low latency, very high throughput large language model inference, and the large NVLink domain is going to be a game changer for that. And so I think people are very comfortable deploying both. And so almost every CSP we're working with are deploying some of both. And so I'm pretty confident that we'll ramp it up just fine.”
-End Quote
My note: having a hard time closing our weaker DLC positions based on what we know is coming. My main regret is not keeping some dry powder to buy at these lower levels.
Margins:
Discussion on margins felt a bit forced as Nvidia’s margins continue to beat the rest of the Mag 7. Consider that Nvidia’s GAAP operating margin this quarter was 62.1%:
Where the market is a tad concerned is gross margins, which peaked at 78.4% and will exit the year in the mid-70% range. We prepped our Members for this in the pre-earnings writeup. The CFO stated: “Your second piece is in terms of our gross margin. We provided gross margin for our Q3. We provided our gross margin on a non-GAAP at about 75%. We'll work with all the different transitions that we're going through, but we do believe we can do that 75% in Q3. We provided that we're still on track for the full year also in the mid-70s or approximately the 75%. So we're going to see some slight difference possibly in Q4, again with our transitions and the different cost structures that we have on our new product introductions.”
Conclusion:
Our pre-earnings writeup expressed concerns about the valuation going into the print. I think the selling after hours reflects the valuation.
Earnings reports are truly 50/50 – nobody can tell you what the market will do following a report. For example, we had high confidence Nvidia would beat, but there’s much more to consider than a beat.
What’s important is to have a strategy, and we do. Our strategy is to trim this position depending on how the price reacts at key levels. Our goal is to then re-allocate what we trim at lower levels. You will get trade alerts to this effect when we think the timing is right.
Damien Robbins, Equity Analyst for the I/O Fund, contributed to this analysis.
This article was originally published on Forbes on Aug 22, 2024,11:24pm EDTForbesForbes on Aug 22, 2024,11:24pm EDT
Bulletproof Nvidia showed an unusual bout of weakness this past month following a report from The Information that Nvidia’s new AI chips are delayed. The report asserts that Nvidia’s upcoming artificial intelligence chips will be “delayed by three months or more due to design flaws,” resulting in a final flush of selling where the stock was down (-15%) in 7 days.
According to the report that was based on two anonymous sources, “if the upcoming AI chips, known as the B100, B200 and GB200, are delayed three months or more, it may prevent some customers from operating large clusters of the chips in their data centers in the first quarter of 2025, as they had planned.” This statement sent the market into a panic as it implies all three Blackwell SKUs will be delayed into the June quarter given the statement a three-month delay may prevent large clusters of Blackwell from not being operable in the first quarter.
It's strange then, to say the least, that according to two of Nvidia’s closest supply partners, there is evidence the GB200s will initially ship in Q4, and are expected to see an increase of production volume in Q1.
The third supplier provides a read-through that the fab producing the chips is not seeing any material impact. This is important as the The Information also asserts the machines fabricating Blackwell GPUs are sitting idle. Per the report: “it is highly unusual to uncover significant design flaws right before mass production. Chip designers typically work with chip makers like TSMC to conduct multiple production test runs and simulations to ensure the viability of the product and a smooth manufacturing process before taking large orders from customers. It’s also uncommon for TSMC, the world’s largest chipmaker, to halt its production lines and go back to the drawing board with a high-profile product that’s so close to mass production, according to two TSMC employees. TSMC has freed up machine capacity in anticipation of the mass production of GB200s but will have to let its machinery sit idle until the snags are fixed.”
The quote above implies the issues were entirely unforeseen, which might not be the case. My firm covers Nvidia’s management team statements quite closely since I first covered the AI thesis in 2018, and management has been quite clear that CoWoS-L packaging for Blackwell will require more time for testing than previous generations. I’ve dug up some of this commentary for you below.
Nvidia is delivering the history’s most aggressive product road map on new fab processes. This is a “move fast, break things” problem, which contrasted to strictly a design flaw, does not mean the architecture inherently has issues. Rather to contrast, the progression of this generation is testing the upper limits of manufacturing complexities. Blackwell with CoWoS-L packaging seeks to increase yields by circumventing a silicon monolithic interposer, and instead, will use an interposer with higher yields to help package the processing and memory components seamlessly together. The result will be to break ground on unprecedented performance gains for memory-intensive tasks.
These nuances matter for tech investors. Around this time, on August 2nd, my firm took the opportunity to buy our last Nvidia tranche at $105.73 in an effort to catch what we believe will be about 25% – 50% upside before price tops out.
We also look more closely at supply chain commentary, as there is one supply chain partner in particular that has reported a mysteriously high level of growth in a segment that is tied to Blackwell. We covered this for our premium members the evening of the supplier’s earnings report on August 6th when Nvidia stock was bottoming at $105.
As a reminder, we don’t make earnings calls, as many factors can affect stock price. Instead, we present quality research so that investors are fully informed to make their own decisions. From there, we take this a step further and publish every single trade we make on our research site. In finance, full transparency is rare, yet through never-ending tenacity, my firm has offered up to 3900% gains on Nvidia alone.
We continue this long-standing dedication to our readers in the analysis below.
Sign up for I/O Fund's free newsletter with gains of up to 2600% because of Nvidia's epic run – Click hereClick hereClick here
TSMC Reports 23.6% MoM Growth in July, Highest in 2024
TSMC releases monthly numbers which would reflect quickly if a highly anticipated release was causing idle machines. Instead, July monthly revenue showed a sharp acceleration from a decline in May and June to a MoM growth of 23.6% to NT$256.95 billion.
Source: I/O Fund
On a MoM/YoY basis, July reported the second largest growth this year:
Source: I/O Fund
TSMC’s MoM growth can be lumpy, yet July month’s 38.3% YoY growth points to a positive start to the September quarter. The company guided for revenue of $22.4 billion to $23.2 billion, representing YoY growth of 31.9% at the midpoint.
Source: I/O Fund
The analyst consensus estimates are trending higher, which typically, you’d see a decline in the analyst estimates on the news of a material delay. Analysts expect Q3 revenue to grow 38.1% YoY to $23.32 billion from the earlier 32.5% growth expected in mid-June and 32.1% growth expected in mid-May.
Note: The analyst estimates below differ slightly from reported figures in the company IR due to the currency conversion. However, we use the estimates below to understand the expected growth rate trend.
Source: Seeking Alpha
TSMC offered positive commentary on its business and raised the outlook when it reported its Q2 report last month. The company’s revenue grew by 32.8% YoY to $20.82 billion and beat the midpoint guide of 27.6% growth, helped by strong AI demand.
On a QoQ basis, the chipmaker’s high-performance computing (HPC) revenues rose 28% QoQ to $10.8 billion and accounted for 52% of Q2 revenue, up from 46% of revenue in Q1. HPC is above the 50% mark for the first time.
Source: I/O Fund
C.C. Wei, Chairman and CEO of the company, said in the Q2 earnings call, “Our business in the second quarter was supported by strong demand for our industry-leading 3-nanometer and 5-nanometer technologies, particularly offset by continuous smartphone seasonality.” There was a similar trend in Q1 as revenues were impacted by smartphone seasonality and offset by HPC revenue.
Wei also said that “over the past three months, we have observed strong AI and high-end smartphone related demand from our customers, as compared to three months ago, leading to increasing overall capacity utilization rate for our leading-edge 3-nanometer and 5-nanometer process technologies in the second half of 2024. Thus, we continue to expect 2024 to be a strong growth year for TSMC.”
They raised the full-year guidance to “slightly above mid-20s percent in US dollar terms” from the earlier “increase by low to mid-20% in U.S. dollar terms.” He further added, “we have such high forecasted demand from AI related business.” Given TSMC has many high-profile customers, the HPC segment alongside the CEO commentary help to differentiate the impact is coming from AI, rather than being mobile-related.
The I/O Fund built a leading AI portfolio beginning with Nvidia’s AI thesis in 2018, with up to 2,600% gains on Nvidia alone provided to our free readers. Premium members receive real-time trade alerts on NVDA and our entire portfolio, including two AI semiconductors we believe are poised for growth with allocations rivaling our NVDA holding.real-time trade alerts on NVDA and our entire portfolio, including two AI semiconductors we believe are poised for growth with allocations rivaling our NVDA holding.
TSMC’s Advanced Packaging
TSMC has limited CoWoS-L capacity to produce Blackwell chips. This is a problem all investors should get comfortable with as we head into 2025.
TSMC’s chip-on-wafer-on-substrate (CoWoS) architecture refers to the 3D stacking of memory and processors modules layer by layer to create chiplets. The architecture leverages through-silicon vias (TSVs) and micro-bumps for shorter interconnect length and reduced power consumption compared to 2D packaging.
There are three types of CoWoS architectures, which replaced multi-chip modules by scaling up the interposer area to fit multiple dies. Current CoWoS interposers are up to TSMC’s 3.3X reticle limit, with the goal of building interposers that can reach 8X the reticle limit by 2027. At the North American Technology Symposium earlier this year, TSMC stated they will reach 5.5X reticle limit by 2025 for more than a 3.5X increase in compute power.
As transistor density increases, advanced packaging solutions help to alleviate bottlenecks by increasing interconnect density, which results in higher signal speed and processing power.
CoWoS-S: this is the most popular CoWoS architecture for GPUs already deployed, including Nvidia’s H100s, H200s and AMD’s MI300s. It uses silicon as the interposer material and is lower cost than CoWoS-R.
CoWoS-R: connects chips with redistribution layers (RDL) wiring as the interposer material, offers InFO technology as an upgrade for HBM memory and SoC integration.
InFO technology reduces the size of components in more powerful devices. By being Fan-Out (FO) instead of Fan-In, TSMC’s process can integrate multiple dies on top of each other with a common I/O connecting layer.
CoWoS-L: combines multiple Si interconnects (LSI) for a reconstituted interposer (RI) that replaces the monolithic silicon interposer in CoWoS-S. By taking the benefits of CoWoS-S, CoWoS-L offers strong system performance while avoiding the yield loss from one Si interposer.
Nvidia designs offer pure ingenuity, for example, the A100s offered sparsity and the H100s offered a transformer engine. We covered the importance of the Transformer Engine for our premium site six months prior to Hopper shipping, which led to entries as low as $10.85 when factoring in the stock split. Ultimately, Nvidia’s design ingenuity combined with TSMC’s process improvements defy Moore’s Law.
Due to TSMC’s CoWoS-L requiring more complexity and precision, it was already expected the validation and testing process would be time consuming. We had stated in the analysis Nvidia Q1 Earnings Preview: Blackwell and The $200B Data Center that “the advanced CoWoS packaging that is needed to combine logic system-on-chip (SoC) with high bandwidth will take longer, and thus, it’s expected that Blackwell will be able to fully ship by Q4 this year or Q1 next year. How management guides for this will be up to them, but commentary should be fairly informative by Q3 time frame.”
Per another source, Trend Force last April: “Although NVIDIA plans to launch products such as the GB200 and B100 in the second half of this year, upstream wafer packaging will need to adopt more complex and high-precision CoWoS-L technology, making the validation and testing process time-consuming. Additionally, more time will be required to optimize the B-series for AI server systems in aspects such as network communication and cooling performance.It is anticipated that the GB200 and B100 products will not see significant production volumes until 4Q24 or 1Q25.”
From the horse’s mouth, Nvidia’s own management team, it was stated during the GTC Financial Analyst Day in March that the very first systems will ship in Q4, but to expect constraints. In a roundabout way, the CEO tells investors what to expect should this happen, which is that customers will continue to build with H100s, H200s and any other supply they can get their hands on.
Atif Malik, Citigroup:
Hi. I am Atif Malik from Citigroup. I have a question for Colette. Colette in your slides, you talked about availability for the Blackwell platform later this year. Can you be more specific? Is that the October quarter or the January quarter? And then on the supply chain, readiness for the new products is the packaging, particularly on the B200 CoWoS-L and how you are getting your supply chain ready for the new products?
Colette Kress:
Yeah, so let me let me start with your second part of the question, talking about the supply-chain readiness. That's something that we've been working well over a year getting ready for these new products coming to market. We feel so privileged to have the partners that work with us in developing out our supply chain. We've continued to work on resiliency and redundancy. But also, you're right, moving into new areas, new areas of CoWoS, new areas of memory, and just a sheer volume of components and complexity of what we're building. So that's well on its way and will be here for when we are ready to launch our products. So there is also a part of our supply chain as we talked earlier today, talking about the partners that will help us with the liquid cooling and the additional partners that will be ready in terms of building out the full of the data center. So this work is a very important part to ease the planning and the processing to put in all of our Blackwell different configurations. Going back to your first part of the question, which is when do we think we're going to come to market? Later this year, late this year, you will start to see our products come to market. Many of our customers that we have already spoken with talked about the designs, talked about the specs, have provided us their demand desires. And that has been very helpful for us to begin our supply chain work, to begin our volumes and what we're going to do. It's very true though that on the onset of the very first one coming to market, there might be constraints until we can meet some of the demand that's put in front of us. Hope that answers the question.
Jensen Huang:
Yeah, That's right. And just remember that Hopper and Blackwell, they're used for people's operations and people need to operate today. And the demand is so great for Hoppers. They — most of our customers have known about Blackwell now for some time, just so you know. Okay, so they've known about Blackwell. They've known about the schedule. They've known about the capabilities for some time. As soon as possible, we try to let people know so they can plan their data centers and notice the Hopper demand doesn't change. And the reason for that is they have an operations they have to serve. They have customers today and they have to run the business today, not next year.
—End Quote
Recently, Nvidia’s VP Ian Buck stated at BofA Financial Conf in June 2024; “So we stated recently in our earnings that Blackwell has now entered into production builds. We started our production.
The samples are now going — will go out this quarter, and we're ramping for production outs later this year. And then everything — that always looks like a hockey stick, you start small and you go pretty quick to the right. And the challenge, of course, is with every new technology transition comes — the value is so high, there's always a mix of a challenge of supply and demand. We experienced that certainly with Hopper. And there'll be similar kinds of supply/demand constraints in the on-ramp of Blackwell certainly at the end of this year and going into next year.”
Taking this full circle, let’s go back to what TSMC said in the most recent earnings call about CoWoS capacity:
Management stated in the earnings call Q&A that the supply is expected to continue to be tight next year, and they are working with OSAT (Outsourced Semiconductor Assembly and Test) partners to increase production capacity.
Gokul Hariharan:
“How do you think about supply demand balance for AI accelerator and CoWoS advanced packaging capacity? And I think in your symposium you talked about 60% CAGR, component growth for CoWoS capacity in the next four, five years. So, could you talk a little bit about how much capacity for CoWoS would you be planning to build next year as well?”
C. C. Wei:
“Gokul, I also try to reach the supply and demand balance, but I cannot today. The demand is so high, I have to work very hard to meet my customers' demand. We continue to increase, I hope sometime in 2025 or 2026 I can reach the balance. You're talking about the CAGR or those kind of increase of the CoWoS capacity. Now it's out of my mind. We continue to increase whatever, wherever, whenever I can. Okay. The supply continues to be very tight, all the way through probably 2025 and I hope it can be eased in 2026. That's today's situation.”
Gokul Hariharan:
“Any thoughts on next year capacity? Are you going to double your capacity again next year for CoWoS?”
C. C. Wei:
“The last time I said that, this year I doubled it, right? More than double. Okay. So next year, if I say double, probably I will answer your question again next year and say more than double. We are working very hard, as I said. Wherever we can, whenever we can.”
—End Quote
My notes: There were many opportunities for TSMC to report a material impact from idle machines – quarterly numbers ending in June, July monthly numbers, commentary during the earnings call from the CEO that establishes the opposite, which is that capacity is primarily the issue (rather than a dire flaw that is halting production) and the company is working hard to increase this capacity.
Earlier this month, TrendForce citing Money DJ’s report, estimated that CoWoS capacity is in short supply at 35,000 to 40,000 wafers this year. With outsourced capacity, 2025 production could be over 65,000 wafers per month.
According to the report, TSMC will assign the orders of the initial stage of CoWoS packaging, Chip on Wafer (CoW) to OSAT partner SPIL. This is the first time the company is outsourcing this process since the demand is high and previously WoS (Wafer-on-Subtrate) process was outsourced while keeping the higher margin CoW process in-house.
According to DigiTimes, the company is expected to have CoWoS production of 60,000 wafers per month in 2025 and a further increase to 70,000 to 80,000 in 2026 after the company recently acquired Innolux Fab. The 2025 production capacity would suggest a 300% increase from 15,000 at the end of 2023.
Every Thursday at 4:30 pm Eastern, the I/O Fund team holds a webinar for premium members to discuss how to navigate the broad market, as well as various stock entries and exits. We offer trade alerts plus an automated hedging signal. The I/O Fund team is one of the only audited portfolios available to individual investors. Learn more here.Learn more hereLearn more here.
Super Micro: The Near-Perfect Proxy for Nvidia
Super Micro stock surged alongside Nvidia over the past year and a half with returns of 659% compared to Nvidia’s returns of 787.8%. Supermicro is a leading partner on building AI systems with Hopper GPUs by leveraging air cooled and liquid thermal designs for AI accelerators to grow upwards of 5X faster than the industry average for subsystems and server systems.
The Hopper generation is primarily air cooled. However, the percentage of air-cooled systems shipped versus liquid cooled systems will change (dramatically) with Blackwell.
In June, we wrote an analysis on AI Power Consumption: Rapidly Becoming Mission Critical which stated that as the industry progresses towards a million-GPU scale, this puts more emphasis on future generations of AI accelerators to focus on power consumption and efficiency while delivering increasing levels of compute. Data centers are expected to adopt liquid cooling technologies to meet the cooling requirements to house these increasingly large GPU clusters.
Specifically, it’s the Blackwell architecture that kicks off the need for liquid cooling. Most servers today are air-cooled yet AI necessitates a shift to liquid cooling as the H100 GPUs are already at 700W of power and Blackwell GPUs will see a 40% increase to 1,000W or higher. The B200 doubles the transistor count compared to the H100 and provides 20 petaflops of AI performance compared to the H100s 4 petaflops. The resulting 3X leap in training performance and 15X leap in inference performance is shifting the focus to liquid cooling as 1,000 watts is too hot to be air cooled.
The B200 systems and chipsets will be the first release to be primarily liquid cooled, according to Dell, who competes with Super Micro on building AI servers. Note that per the statement from Dell in March, the B200s are due in early 2025.
Tom’s Hardware has also stated that direct liquid cooling will start with Blackwell: “Even Nvidia's high-end H100 and H200 graphics cards work well enough under air cooling, so the impetus to switch to liquid hasn't been that great. However, as Nvidia's upcoming Blackwell GPUs are said by Dell to consume up to 1,000 watts, liquid cooling may be required.”
VP Ian Buck of BofA GTC Conference in June of 2024 also stated: “The opportunity here is to help [customers] get the maximum performance through a fixed megawatt data center and at the best possible cost and optimized for cost. By doing 72 GPUs in a single rack, we need to move to liquid cooling. We want to make sure we had the higher density, higher power rack, but the benefit is that we can do all 72 in one NVLink domain.”
Super Micro is a proxy for Nvidia as its growth has been in lock-step with the AI GPU juggernaut since the launch of the H100 nearly two years ago. Most importantly, we have a key metric from Super Micro that is specifically tied to Nvidia’s Blackwell launch, which is the ramp of liquid cooling.
Liquid cooling has been around for 30 years, yet the H100s and H200s launched with air cooled systems. Today, Super Micro builds HGX AI supercomputers with racks that support 64 H100s, H200s or B200s with direct liquid cooling (DLC), saving up to 40% of energy costs. Although H100s and H200s have the option for DLC, the CFO of Super Micro has stated that as GPUs and CPUs run over 1,000 watts, the benefits of liquid cooling are “going to start to become painfully obvious.”
Per the CEO in last month’s earnings call, it was the months of June and July specifically when DLC started to ramp: “I mean as you know liquid cooling have been in the market for 30 years and market share compared with overall datacenter size always small, less than 1% or close to 1%, I would have to say. But just June and July two months alone, we shipped more than 1,000 racks to the market. And if you calculate 1,000 racks, AI rack is about more than 15% on a global datacenter new deployment.”
Next quarter will mark the highest quarter growth in Supermicro’s history with guidance for 206.6%, an acceleration from the previous quarter’s growth of 144%. This is 590 bps higher than Super Micro’s previous record quarter for 200.7% growth.
Source: I/O Fund
Direct Liquid Cooling is Surging
Considering that Blackwell is a clear catalyst for direct liquid cooling, it is odd to say the leastthat Supermicro reported on August 6th that demand for direct liquid cooling is surging, a mere four days after The Information’s dire report.
According to Super Micro’s earnings report, the company’s direct liquid cooling capacity grew 50% month-over-month from 1,000 racks per month to 1,500 racks per month. By year end, the company will grow to 3,000 racks per month, resulting in 200% growth in six months.
This represents an increase from Super Micro’s original estimate the company would end the year with 1,500 racks. The CFO stated: “But even we were surprised by the acceleration that we saw in the liquid-cooled rack market.”
SuperMicro offers liquid cooled H200 HGX systems, yet the H200s run up to 700W; not the 1000W that necessitates DLC. I have yet to see where the H200 was expected to drive overnight demand for DLC, rather, it’s been expected for some time that Blackwell would be the catalyst for the DLC market.
To put the sudden surge in context, Super Micro stated: “I believe for June and July in last next two months we may ship at least 70% to 80% of liquid cooling compared with all the liquid cooling in the world. So for liquid cooling, we have at least 70% to 80% market share” – the readthrough is the DLC market skyrocketed very suddenly in the last two months.
Supermicro’s report is communicating that servers that require direct liquid cooling are soaring (suddenly) as of June and July from 1% of all new servers shipped to 15% at 1,000 racks. Management is also communicating that it’s expected to continue to soar to 3,000 racks by the end of this year, reaching up to 30% of servers shipped.
Yet if Blackwell is materially delayed, how can liquid cooling be skyrocketing?
A Few Theories I’m Working With:
Theory #1: The Delay Was Accounted For:
Per the GTC commentary from management, the very first GB200 systems will ship in Q4 and will ramp from there, with the understanding Blackwell will be capacity constrained. Financial analysts knew CoWoS-L could present delays, and the April press release from Trend Force clearly describes this, stating CoWoS-L is “making the validation and testing process time-consuming.”
Nvidia reiterated that Q4 is when the first systems would ship after The Information’s report with an Nvidia spokesperson stating to The Verge: “Nvidia expects production of the [B200] chip “to ramp in 2H,” according to a statement that Nvidia spokesperson John Rizzo shared with The Verge. “Beyond that, we don’t comment on rumors.”
The delay may have already been accounted for, as discussed, it’s a new packaging process and a more complex chip, with many statements on record it would require additional testing. This would help explain why TSMC and Super Micro are raising/beating estimates driven by their AI segment as it implies their guidance was aligned with the delay.
Theory #2: The GB200s NVL36 and NVL72s are Hogging CoWoS-L Capacity
My firm has been reporting on X for months that GB200 demand is surging. For example, UBS said that it believes “demand momentum for $NVDA Blackwell rack-scale systems remains exceedingly robust” and that the “order pipeline for (Nvidia's) NVL72/36 systems is materially larger than just two months ago.”
According to reports from Wccftech: “Team Green is expected to ship 60,000 to 70,000 units of NVIDIA's GB200 AI servers, and given that one server is reported to cost around $2 million to $3 million per unit, this means that Team Green will bag in around a whopping $210 billion from just Blackwell servers along, that too in a year.”
The weight of that report cannot be overstated as it implies 26% upside to 2025’s estimates based on one SKU alone. In fact, this one SKU is expected to drive 9% more revenue than analysts currently have estimated two years out for FY2027.
Source: Seeking Alpha
Theoretically, if the GB200 systems are seeing enough demand to exceed FY2027 estimates (per the preliminary data), Nvidia would be wise to cancel the B100s and B200s built on CoWoS-L capacity entirely, and switch these SKUs back to CoWoS-S. There’s a write-up on new SKUs based on CoWoS-S capacity and air-cooling from Semi Analysis here.
Here’s why the GB200 can drive this kind of revenue so quickly:
Nvidia’s GB200, featuring one Grace CPU and two B200 GPUs, is estimated to sell for ~$60,000 to $70,000.
In the NVL36 configuration, featuring 18 GB200s (18 Grace CPUs and 36 B200s), each GB200 would be selling for $100,000 at the current estimated ASP of $1.8 million.
In the NVL72 configuration, featuring 36 GB200s (36 Grace CPUs and 72 B200s), each GB200 would be selling for ~$83,333 at the current estimated ASP of $3 million.
In this case, Nvidia would theoretically prioritize the GB200 NVL36 and NVL72 as the price points are quite high. The two NVL36 and NVL72 rack configurations carry a ~27% to ~54% higher selling price per GB200, making it understandable why Nvidia would focus on the racks given production constraints from CoWoS capacity.
Ultimately, reconfiguring lower priced SKUs will not matter to Wall Street if it’s based on outsized demand for GB200s. This theory hinges on Super Micro’s report, as it’s the sudden surge in direct liquid cooling sales that is truly mysterious. From my vantage point today, it feels nearly impossible that Super Micro could report this level of surge in direct liquid cooling from 1% of systems in May, to 15% of systems today, to 30% of systems by the end of the year and for there to be a material, unforeseen delay in every Blackwell SKU.
If the B100s and B200s are pushed out in favor of the GB200NVLs, then next year will be game-on for Nvidia investors as these systems sell at a high multiple. Keep an eye out for where bad news now (some GPUs are canceled) eventually becomes good news over the next four quarters (in favor of systems priced 36X to 72X higher).
Foxconn Earnings Call Commentary:
Briefly, I’d like to mention Foxconn has recently stated in an earnings call: "We are on track to develop and prepare the manufacturing of the new [Nvidia] AI server to start shipping in small volumes in the last quarter of 2024, and increase the production volume in the first quarter of next year.”
The company also indirectly debunked The Information’s assertion that “it is highly unusual to uncover significant design flaws right before mass production” with Foxconn stating the opposite "It is normal to dynamically adjust [shipment schedules] when the specs and technologies of a new product are largely upgraded. Whether the shipping schedule changes or not, Foxconn will be the first supplier to ship the first batch of GB200," Wu said.
Note Foxconn specifically calls out shipping the GB200, rather than the B100, which was due to ship first. Hopefully, by now it’s clear to our readers should the B100 be bumped, this could have a bullish readthrough if Nvidia re-allocates CoWoS-L capacity to the higher priced GB200 systems.
The H200 is a Force of Its Own
To further the conversation on why a delay in the B100s and B200s can be absorbed, it’s worth taking a moment to discuss the H200.
The H200 is shipping now and is a force of its own with 141 GB of HBM3e memory, up from 80 GB of HBM3 memory in the H100. The GH200 superchip is also equipped with HBM3e and is shipping this quarter.
By significantly boosting memory per GPU – up ~75% from 80 GB of HBM3 in the H100s – the H200 allows Nvidia’s customers to address memory-constrained workloads, such as workloads requiring the largest LLMs, which were built and trained on the H100s. This will fill the gap between shipments of the H100 and Blackwell by easing one critical bottleneck to AI training – memory bandwidth.
The question that I’ve seen raised time and again by investors is why is Nvidia’s GPU demand is this durable in a typically cyclical industry? The answer lies within the H200 and Blackwell. As VP Ian Buck explained at BofA’s GTC Conference in June, “From the end of '22 to today, I think we've improved Hopper's inference performance by 3x. So we're continuously making the infrastructure more efficient, faster and more usable. And that gives the customers who have to now buy at a faster clip, confidence that the infrastructure that they've invested in is going to continue to return on value and does so.”
More importantly, Buck emphasized that hyperscalers “can retire their old legacy systems that maybe they've just left, not upgraded. They can accelerate the decommission of the older CPU infrastructure.” Essentially, Nvidia’s customers can free up megawatts of power and hundreds of racks (and save millions with performance and efficiency gains providing lower TCOs) by decommissioning prior generations of GPUs or CPU-based servers, and this goes for the H200 and Blackwell. Customers can retire older GPU generations such as Volta and Ampere and refit it with H200s, while waiting for Blackwell chips to build new infrastructure, allowing them to benefit from the memory upgrades while in mid-cycle for the Blackwell upgrade.
On the HBM3 side, Micron, SK Hynix and Samsung are intertwined in a deep competition for supply, with SK Hynix serving as the primary supplier for the H100 and Micron being the first to announce itself as the supplier for the H200. Micron has said it is sold out of HBM3e supply through 2025, with preparations and discussions already being made for HBM4 and HBM4e in 2026 and beyond. SK Hynix also revealed earlier in May this year that it was nearly entirely sold out of HBM through 2025. On the other hand, Samsung has reportedly struggled for some time to validate its HBM3e chips with Nvidia due to power consumption and heat issues.
We’re still seeing no signs of slowing for H100 and H200 demand, with DigiTimes reporting last week that H100 and H200 production volumes have been “increasing monthly.” There are also signs in the broader DRAM market that point to HBM demand remaining robust, another signal pointing to lasting Hopper demand. DRAM revenue in the June quarter surged nearly 25% QoQ to $22.9 billion, driven primarily by HBM demand and rising prices due to “aggressive procurement strategies” from buyers.
Conclusion:
As of now, there’s a disconnect between next fiscal year’s revenue estimates of $167 billion and the $210 billion in GB200s alone expected to ship next year. Perhaps analysts are waiting for signals the supply chain can produce these outsized orders. So far, so good with the signals we see from TSMC and SMCI’s most recent earnings reports. Foxconn commentary helps, as well.
Where Nvidia investors run a risk is the valuation of 25X forward PS and 45X PE Ratio as it’s the highest the stock has traded since the market has priced in the AI accelerator boom. My firm believes in an active approach to managing risk. For example, if you had bought the 2022 top in Nvidia, you’d currently be up over 275%. If you bought the October 2022 low, you’d be up over 1100%. It is unlikely many bought the top and bottom in any stock (we actually did buy Nvidia at the very low on October 18th, 2022, but it’s rare). Yet, being cognizant of the larger trend and pattern in play has allowed us to increase our return while decreasing the risk with Nvidia.
Point being, we actively seek to buy quality companies at lower prices. Let the market (with help from the media) doubt the AI juggernaut in its first inning, let them drag the price down, and then our plan is to pounce … because Blackwell is on its way, the GB200s are going to crush expectations in FY2026, we are getting the green light from suppliers the delay is immaterial at this time, demand/big tech capex remains high, and let’s be real, nothing can stop what’s coming.
Please note: The I/O Fund conducts research and draws conclusions for the company’s portfolio. We then share that information with our readers and offer real-time trade notifications. This is not a guarantee of a stock’s performance and it is not financial advice. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis. Beth Kindig and the I/O Fund own shares in NVDA at the time of writing and may own stocks pictured in the charts.
