A few notes before we look at Palantir:
- We are looking to close Atomera and will look to get back in when the timing is better. We feel the message in the earnings call is that during supply shortages, things are moving slowly for Atomera. Meanwhile, small caps are looking like they could break out (follow Knox on the forum and on webinars) and we prefer to put our money elsewhere for now. Atomera could still move but we have to make these decisions to keep the portfolio at a reasonable size.
- There’s a chance we re-enter Vuzix as small caps are starting to break out. Knox might be seeing a setup he wants to take and the company had some good quarters in the past. Note: this is likely good news for many small caps, not only the ones we own.
- Confluent is a strong company. We stepped aside until the lockup expires as a matter of discipline. Look for us to put MongoDB in the LTBH portfolio and Confluent at some point, as well. We are encouraged by cloud results so far this earnings season (so far, so good).
Palantir Analysis:
We break down Palantir’s product below and we believe the Apollo layer is especially interesting, competitively speaking. We also point out that government contracts will likely boost the company’s revenue in the near term. The Obama Administration used Palantir for many government projects and we believe the Biden Administration is a tailwind for the company in the near-term. With a company that is two-thirds deal value from the government, this piece cannot be ignored.
The company has as many risks as opportunities. We go through those risks below, mainly the price of the product, the unusually high stock based compensation, widespread ethical concerns (for 10+ years), and more agile AI/ML competitors sprouting up to compete for commercialized accounts. Due to these risks, we may not hold for the 3-5 year time frame that we typically target, rather are entering as a LTBH for the 6-month to 1-year tailwinds that we are expecting from increased government spending. This is distinguished from momentum positions that are often higher beta and/or moving in price. In this case, we think Palantir being off 50% from its all-time high of 39.00 does not reflect the current tailwinds from the government segment. We are encouraged by the commercial growth, as well, but it’s the government spending in the near-term that could cause a material change to the story. The upcoming earnings report will tell us more.
Palantir has two platforms: Gotham and Foundry. There is a layer between the two platforms and applications called Apollo. The Apollo layer is where innovation has been rapidly occurring and helps contribute to Palantir’s competitive edge (more on this below).
Gotham and Foundry create a unified data set for actionable insights across industries such as manufacturing, product development, and customer experience. The data that Palantir gets is from the customer database although the company may use other data sets for government customers, such as scraping social media or other publicly available information on the web. The traditional deployment includes hosting Palantir’s servers in a customer’s data center.
The difference between Palantir and competitors, such as Tableau, Alteryx or Cloudera is that Palantir is able to answer questions a model cannot answer. Traditional business intelligence companies require a complete data set whereas Palantir is able to tackle situations where there is not a complete data set.
Palantir Gotham was the company’s first platform, built for government operatives in defense and intelligence sectors. The platform enables users to identify patterns hidden deep within datasets using semantic, temporal, geospatial and full-text analysis.
The Graph product allows data to be seen as nodes and edges to visualize and plot characteristics in a logical manner.
Map brings geospatial capabilities to track geo-located objects and events and to create heatmaps for the density of the objects.
Object Explorer is powered by the Horizon in-memory database, which competes with Apache Spark for letting users query billions of objects. The data provides further analysis for Map and Graph data.
Browser: This enables search queries for investigations and surfaces information, runs relevant searches, displays key data points and answers analytical questions.
Palantir Foundry is the commercial offering and has four layers of tooling: Foundry Core, Data Foundation, Ontology and Workflows.
This four-step process does the following:
- brings volumes of data into one place,
- transforms the data into a format that analysts can work with and enables validation in a number of programming languages
- the “ontology layer” allows datasets to be turned into real-world concepts
- workflows is where it all comes together in an integrated environment for object exploration, point-and-click top down analysis, code authoring, time series analysis, data science and application development. When a user has a question, it answers it using all layers and tools available.
Palantir describes Gotham and Foundry as the “ability to construct a model of the real world from countless data points.” Unlike a SQL database, natural language is used to query data and return results in real-time rather than through strings.
Apollo is the Linchpin:
The company has a third platform or layer called Apollo and also Apollo for Edge AI. This product provides continuous delivery and an automated configuration layer that allows Foundry and Gotham to work across all cloud environments and also in places where there is little to no connectivity. On top of Palantir being able to form conclusions from incomplete data sets, the company can also deploy its platform and applications anywhere.
Palantir’s marketing team says Apollo “goes where no SaaS has gone before” because it allows what is done on-premise to also run on multi-cloud SaaS with code that is deployed across all environments rather than written for a specific environment. The orchestration allows for on-hardware AI models to consume real-time data from sensors, radio, geo-data and time series data. Where bandwidth is not an issue, the company transmits all raw inputs and enriched metadata from models. Where there are constraints, the platform transmits meta-data only which can reduce bitrate by 20X. At times, a simulated environment can be created with Palantir’s Edge AI from historical data to help train AI models. The simulated environment is then deployed at the edge. With Apollo, Palantir’s centralized operations team is capable of 41,000 updates per week at no additional cost.
Apollo Edge AI links together satellites to lower latency for the AI-enabled decision chain by orchestrating up to 237 satellites in what the company is calling a “meta-constellation.” This meta-constellation optimizes hundreds of orbital sensors and AI models to power Palantir’s models. One example they provide is tracking submarines that pose a threat to the U.S. and its allies. In this case, submarines are being tracked on a granular level in areas where there is no bandwidth available. These are the kinds of obstacles that Palantir overcomes while being independent of one cloud environment, such as AWS or Azure.
Financials:
Palantir is growing its annual revenue of roughly $1 billion by 50% for estimates of $1.5 billion in 2021. This looks like it will be accomplished with the last two quarters at revenue growth of 49% year-over-year growth. Current estimates for Palantir in the upcoming quarter are at $386.53 million, or growth of 33.5%. To me, these estimates seem low considering the commercial growth the company has been posting. On top of revenue > $1 billion and growth > 40%, Palantir is free cash flow positive with a 13% adjusted free cash flow margin and adjusted EPS of $0.04.
The main metric for Palantir is commercial revenue, which has accelerated nicely over the past few quarters. In the last quarter, commercial revenue grew 90% year-over-year. The company is also adding commercial customers faster than overall customers at 32% compared to 13% for total customers. In the quarter ending in March, the company reported revenue growth of 72% year-over-year. This was slightly lower than government revenue growth of 83% YoY.
The company also grew total contract value booked from $337 million to $925 million, although this is a mix of both government and commercial contracts. According to the fine print, the maximum potential revenue from commercial contracts is $348 million and an additional $195 million in commercial contracts that are subject to negotiation and approval.
The deal value also increased 63% to $3.4 billion, however, of this $428 million comes from commercial contracts and $195 million comes from commercial contracts currently under negotiation. Therefore, the majority of the deal value increase came from the government.
For FY 2021, the company plans to double its adjusted free cash flow in the upcoming quarter from $150 million to $300 million.
Stock Based Compensation:
Palantir has some of the highest rates of SBC in the cloud universe. Over the last twelve months, SBC was 114% of sales, which is well above the peer median of ~18%. The issuance of SBC is dilutive to shareholders and can weigh on the share price in the near term. However, Palantir recently completed its IPO, which is typically a period of outsized stock-based compensation.
Looking forward, Palantir’s rate of SBC will likely normalize to a more sustainable rate, which will lessen the impact of dilution and should benefit shareholders going forward.
A key benefit of high SBC is that employees become owners in the company and have a vested interest in the company’s success, which can even help reduce turnover and improve productivity. The biggest concern Bradley sees with high rates of SBC is if the SBC is repurchased via stock buybacks but is still excluded from adjusted EBITDA and earnings. This accounting trick can cosmetically improve the presentation of profitability by excluding payroll expense from non-GAAP metrics. However, Palantir does not appear to be playing these games, as it has not repurchased any stock during the year.
Catalysts:
Some real-world uses for Palantir include Hershey’s using the software for global food distribution and to correlate weather patterns with snack consumption. Chase Bank and other financial firms have used Palantir’s data analysis to identify troubled properties and ensure employees are not committing fraud (and in turn, the management team was actually spied on instead).
Pharmaceutical companies use Palantir to expedite the development of new drugs – this being a substantial use case during Covid and partly why Palantir’s revenue has accelerated. In the last earnings report, Palantir discussed companies leveraging Palantir’s software for R&D and manufacturing to accelerate development. The software helps health care data be shared to share trial data.
Utility companies use Palantir to monitor equipment, such as to monitor equipment in mining shafts or for grid management and safety. The powerlines from PG&E in California created wildfires and experience ongoing power outages during heavy winds. PG&E partnered with Palantir in early 2021 to help assess where the most danger is for power shutoffs and for wildfire risk assessments.
Climate change initiatives coming from the government will also be a tailwind for Palantir as the company’s software is used to help companies de-carbonize and achieve low carbon footprints. The more spent here, the more Palantir will see additional tailwinds.
DataRobot is a popular company used for unifying data for AI and they are partnered with Palantir to help forecast demand.
As stated, Palantir was first hired by Obama for border patrol with the New York Times reporting “Palantir’s technology was used extensively by the Obama Administration.” It is not clear as to whether the Trump Administration used Palantir or if the agencies, such as the FBI and CIA did during the years the Trump Administration was in Office. In other words, I am not sure if Palantir is bipartisan or not but my understanding is the company saw more government contracts during Obama and now Biden. We also saw Biden place a former Palantir advisor as the director of national intelligence. The DNC is headquartered in Denver and the company recently moved to this city. Alex Karp attempted to state sensational reasons for this move, which I called out as simply creating headlines. I believe the move was strategic for Palantir to be closer to the money.
Risks:
The closest competitor for Palantir is Semantic AI, which supplies graph-based analytical platforms to the DoD and other government agencies. There will likely be more competitors in the near future as the AI/ML market is built out. For instance, there is energy-specific software such as Stem that uses AI software to optimize energy resources and battery usage by using algorithms to issue forecasts that then work across the grid, batteries and solar for optimal output. Stem claims to have taken over 100 energy storage systems that were previously managed by competitors and is also used across Big Tech, such as Facebook, Amazon, Apple and Home Depot. In this case, one could argue Stem serves the commercial market whereas Palantir is more suited for larger utility companies due to its government-sized solution. Essentially, the risk is that Palantir could be “too much product” for commercialized companies that prefer a simple solution. You can read our analysis regarding Stem here.
Tiberius is a database used for administering the Covid vaccine. USA Today reported complaints from a few healthcare agencies that Tiberius was often wrong and did not improve results compared to their own in-house databases.
Palantir greatly centralizes datasets and AI/ML — which is a risk. You’ve likely read my analysis on the Blockchain is Going to Eat the Internet and why decentralization is important. Using Palantir for defense is one thing, but now that Palantir is beginning to move into other industries, the blurring of the lines as to where the government ends and the free market begins is problematic with a company like Palantir. Palantir’s greater loyalty will be with the government (it’s biggest customer) yet their software is now inside company databases. The United States tends to prefer a separation across government bodies whether it’s church/state or state/federal or judicial/executive/legislative, when possible. What Palantir is proposing is that a heavily government-funded company be the middleman.
What affect could this have? Already, Palantir has been used to hunt down illegal immigrants and to enter their homes for arrests. This article is worth a read for more information. Uber has been in a string of never-ending lawsuits over the independent contractor/employee debates – another human rights issue that a tech company faced. These lawsuits threaten Uber’s business model, and even after getting the measure on a ballot which passed in California, the class action lawsuits are still ongoing after the State of California decided to sue the company. We predicted this would be troublesome long-term for Uber as part of our bear thesis at the time of IPO. As Palantir moves outside the government, I expect we could see some States and non-profits fight the company on the use of its software. There is a history of non-profits, such as Amnesty International, calling out Palantir on how the software is used in terms of targeting specific individuals. The bigger Palantir gets, the more the public and critics will see how powerful (and invasive) the software can be. As of now, Palantir has chosen to target illegal immigrants who can’t bring a class action lawsuit – hence non-profits stepping in. If the company were to target United States citizens, I would fully expect lawsuits to pop up.
To help illustrate, the week Palantir went public, Hootsuite stated the company would terminate its ICE contract due to disagreements within the company. The CEO of Hootsuite tweeted: “We typically do not make public facing statements about specific customers or contracts. However, due to the attention around this particular case we can confirm that Hootsuite has decided not to do business with the U.S. Immigration and Customs Enforcement.” Tech companies often see employees engage in protests when a company contracts with the government on AI-driven war missions and privacy issues.
In the past, Google ended a contract with the Pentagon when employees protested using AI for lethal purposes. Karp became controversial and challenged Google on this decision, saying it was a “loser” position. Palantir could become subject to competing for talent with companies that are more privacy-compliant or viewed as being more ethical. Here’s an example about how they describe their hiring process: “We spend time thinking about exactly what gamma radiation your incoming Bruce Banner needs to turn into the Incredible Hulk. And then we irradiate them.” The hyperbolic description of using “gamma radiation” is likely just the stock-based compensation.
Conclusion:
The stars (and satellites) are aligning for Palantir, and with government spending, it has the ingredients to become a stock market darling if the revenue accelerates. The product is often framed as captivating and the company will likely sell Wall Street on commercial growth. Regardless, the ethical issues can mire the company long-term, and at its core, Palantir is still a government contractor. We are more likely to be 3-10 year bulls for decentralized blockchain companies that handle data in privacy compliant ways over a heavily centralized company. However, for the sake of the current tailwinds, we have entered the stock and added it to our LTBH portfolio.
