Analyzing Michael Burry’s Palantir Short Report

When Michael Burry publishes a short report about one of the hottest stocks on Wall Street, everyone wants to read it. We’ve analyzed dozens of short reports over the years and none have been behind a paywall. But that’s where Burry’s work sits, shielded from those not willing to part with $39, the minimum it requires to read this 10,000-page manifesto. So, we raided the office slush fund and read every one of those 10,000 words so you don’t have to. The best part? We’re going to do something that few seem capable of doing. We analyze Burry’s points of contention objectively in a concise manner.

Swingers have, “the lifestyle.” Stock cults have “the community,” a loosely knit group of people who orbit around a set of key characters that pump a cult stock relentlessly. They’ll drool over every press release, proudly talk about all the merch they’re being sent, refer to the company as “we,” and generally just make a complete mockery out of responsible investing. No best practices are preached. It’s all about spending every dime on the sacred cow. And the worst side effect of this behavior is irrationality.

Every bit of academic evidence conclusively points to stock picking being a really bad idea. If you decide to go down that path anyway, any criticism of the sacred cow you’ve hung your entire net worth on will be seen as a big scary threat. That’s why you see so many Palantir instant analysts responding with 10,000-word diatribes to Burry and challenging his integrity. (Unlike his many critics, Burry has relevant industry experience.) Just like the Streisand Effect, these critics are only making people more interested in Burry’s viewpoints. Turns out his 10,000 word manifesto is insightful, interesting, but largely unactionable.

Burry’s 10,000 Word Manifesto

The introduction starts with a pretty fair assessment of what Palantir $PLTR does, certainly one more precise than the verbose explanations you’ll come across in the wild. He then questions why a company full of geniuses took so long to start making money, and calls out some instances in the past when they may have used false marketing to protect their image. It’s no different from when we called out Palantir for sweeping their net retention rate (NRR) metric under the carpet when it fell to 107%. Since then, it’s risen to 139%. (More on this in a bit.) The manifesto goes on to describe Karp’s quirkiness, the company’s propensity to blow through cash, and keeps citing the book The Philosopher in the Valley along with some anecdotal source whose contributions Burry can’t verify. (You see anecdotal evidence presented a lot in short reports and we largely ignore it.) Ten pages in we finally encounter something interesting.

The R&D Spending Problem

There has always been some confusion around Palantir’s “forward deployed engineers” or FDEs who are no different from Salesforce consultants that work onsite when you’re deploying a global CRM system that takes the better part of a year. For sophisticated enterprise software implementations, there will be lots of traditional consulting happening. Instead of breaking this work out into a “services” bucket, Palantir classifies this work as “R&D” which one could argue distorts profitability. This begs the question – just how much is Palantir spending on R&D as a percentage of revenue? About 14%, a number that’s declined from 28% in 2021. Compare that to your average SaaS firms which spends about 25% of revenues on R&D and this seems like a nonissue.

Whether Palantir’s engineers are doing consulting or developing new versions of the product, it’s an appropriate rate of spending that’s declining over time. And what exactly changes if you stuff R&D spending into cost of goods sold (COGS)? Well, you’d have lower gross margins which might command a lower valuation multiple. (More on this in a bit.)

The SPAC Debacle

We’re only going to mention this point of contention because it’s something we addressed years ago. Palantir’s arrangement with SPACs was something we disapproved of from the very beginning because we identified them as vehicles that would do nothing but destroy shareholder value, and that’s exactly what happened.

Basically, Palantir sold their solution to many of these SPACs in exchange for equity instead of dollars. Much of that equity turned out to be worth far less than what it was issued at, and Palantir had no right to start speculating on volatile assets instead of just taking the cash (hint: it’s probably because these deals wouldn’t have happened without these arrangements). At one point they also held a speculative gold position, something that shareholders aren’t paying them to do. It’s like when a stagnant software business decides to put bitcoin on their balance sheet (cough, MicroStrategy, cough). While the SPAC situation turned out to be less than ideal for Palantir, it hasn’t had a big effect on their ability to sell their enterprise AI platform. But is it really “AI?”

Palantir In Not an AI Company

You know how to tell someone is a Palantir shareholder? They’ll tell you. And you can drive most of these characters mental by saying one simple sentence. “Palantir is not an AI company.” Before you know it, an army of fintwat prognosticators will be attacking you with 5,000-word essays about how YOU JUST DON’T UNDERSTAND PALANTIR! Newsflash, folks. It doesn’t matter what Palantir does when their revenue growth looks like this:

We’ve said the same mantra over and over since we first started covering Palantir and AI heavily nearly a decade ago. The ground truth to show that you’re “selling AI” is revenue growth. Plain and simple. Palantir’s revenue growth is impressive regardless of whether the company has developed their own LLM or not. They haven’t, and don’t need to. They just put a wrapper around your organization, extract the data and apply meaning to it (muh ontology), and then feed it to the LLM du jour. If someone doesn’t like using Grok because Elon Musk is Satan, fine. They’ll use Gemini.

Burry’s critique of Palantir not being an AI company points to his desire to stir up the hornet’s nest a bit. He’s probably going to run this playbook again with similar results too. (Cough, ASTS, cough.) Clever marketing move, but again, not something investors should waste time caring about.

Problems With LLMs

For someone with a strong background in investing, it seems like Burry wants to really understand the technical risks such as LLMs hallucinating. He postulates that Palantir’s quickly coded wrapper simply passes through the LLM interpretations and contains no mechanism for monitoring how well LLMs function.

Palantir brags about how fast they onboard customers at their AIP demos, but these companies will soon realize it’s not as easy as the brown-bag demo implied, says Burry. He also believes Palantir’s wrappers might easily be replaced in the future, and that “bespoke failure mitigation will continue to be necessary at great expense.” In other words, that R&D spend will continue to increase. Except that it’s not, and customers aren’t canceling. Since Palantir’s NRR is now reflecting their company in a favorable light, they’re stopped tucking it away in a footnote.

If, as Burry claims, customers are running into problems with Palantir’s solution, then we would expect this to be reflected in a declining NRR and gross retention rate (GRR).

The Rapid Rise in Net Retention

Palantir doesn’t provide GRR which means a very strong NRR could mask underlying churn. Indeed, we see Palantir hitting a record NRR of 139% which is well above the average 120% you’d expect from a quality SaaS firm. Burry provides an interesting possibility for the explosive increase in NRR that Palantir realized over last year. It’s because of all those AIP bootcamps.

Customers who landed small via bootcamps at $500,000 and expanded to $2 million or more within twelve months show 300% or higher individual retention on a trivially low base, mechanically inflating the blended figure.

He proposes that the decline in NRR was reversed artificially by signing up customers at lower contract amounts, and then much higher amounts as shown in the above example. This distortion is masking customers who have been on the platform for a while and then say “nah, it’s not for us” and cancel. He postulates that it often takes a while for a company to admit its mistakes, so these cancels will start showing up in the next several years. Burry argues Palantir’s customers will soon realize they’re paying too much for something that can be replicated internally, especially with the advancements we’re seeing in AI. Palantir could very easily start reporting GRR to alleviate these concerns. If it’s true they have hostages, not customers, then this simple metric reinforces that.

Always identify key metrics to monitor that will prove a theory is coming to fruition. For Palantir, everything comes down to continuing that strong revenue growth. Profitability isn’t a huge concern this early on. The more they scale, the more they can afford thinner margins and still be highly profitable.

International Expansion

Burry correctly points to concerns around how Palantir’s vocal and controversial leader, Alex Karp, has sidled up to the Trump administration which – some believe – is contributing to Palantir’s inability to expand overseas. Karp describes this as “a bandwidth problem” to which Burry quickly shouts, “Ah hah! You’re a consultancy!” Not necessarily. A “land and expand” business model that requires “forward deployed engineers” to help implement is no different to how Salesforce implementations require “Salesforce consultants” to be onsite for long periods of time helping companies with migrations and implementations.

As for Karp’s willingness to play politics, it’s a double-edged sword. The friendships he’s forming in the current administration may come back to haunt him during the next administration. You can hear the rumblings from afar. “Palantir is an evil company that the current administration uses for surveillance.” What if those murmurings become more vocal when the next administration comes into power? This is why Palantir’s reliance on the U.S. government is a concern we’ve been persistently raising over time.

Commercial vs Government

We’ve always viewed Palantir’s reliance on the U.S. government and the CEO’s willingness to engage in politics as big red flags. This became even more of a concern when Elon Musk revealed how his DOGE efforts threatened to cut a huge swath of government spending which very well could have affected Palantir. However, the numbers tell us this has been a nonissue. Government growth has persisted strongly which leads to another problem. When will we see Palantir’s commercial revenues command a majority of total revenues?

Progress has been slow with commercial revenues slowly edging up over the years to where they sit today – 48% of total revenues. They’re not a majority, but are getting close. Is this acceptable? Perhaps, when you consider that government revenues have kept growing strongly despite one of the biggest axes that will probably ever be swung on Capital Hill – DOGE.

If Palantir can weather that turmoil, even thrive despite of it, then perhaps their strong reliance on the U.S. government isn’t that big of an issue. The next milestone for this risk will be three years from now when the U.S. welcomes a new administration, but no need to wait until then to invest. The ground truth is always revenue growth, and Palantir has that in spades. So why not go long?

Palantir’s Valuation

This brings us to Burry’s most important point of contention, in our eyes, which is valuation. Instead of using valuation ratios, he provides his own supporting evidence that the stock is worth something like $46 a share. His methods may be different, but the premise is the same. When a company’s valuation reaches stratospheric levels – as Palantir’s has – then one of two things will happen. Either fundamentals will catch up, or the valuation will revert to the mean. Below you can see how Palantir’s simple valuation ratio has evolved over time.

Our simple valuation ratio takes market cap and divides it by annualized revenues, while your typical price to sales ratio uses trailing 12-month revenues. That allows us to ascribe a “cheaper” valuation to companies that are growing more quickly. We don’t invest in any company whose shares trade at three times our catalog average which is around 7.6. That means Palantir would need to trade at an SVR of 22.8 – or $53.47 per share – before the valuation would be at an acceptable level to consider investing in.

Some might have a higher valuation cutoff, and that’s fine. But Palantir’s continuing nosebleed valuations are – at least in some part – caused by Karp’s focused messaging which caters to the broad swath of retail investors, many of which spend all their time cheerleading the company. Those who take the slightest critical approach towards analyzing Palantir are labeled as evil shorts by Karp which is a red flag in itself. Like Musk, he is a double-edged sword for his company.

Conclusion

Burry’s manifesto was a well-written story that was interesting enough to read, but not interesting enough to merit paying for, or dropping everything and writing lengthy rebuttals. That’s exactly what all the Palantir perma-pundits did, and the drama they created worked perfectly in Burry’s favor. He received the attention/subscriptions/influence he was looking for.

Burry showed us he has the chops to take down companies that really have problems as opposed to one of the most popular stocks today which is showing tremendous growth acceleration. The best thing that could come out of this would be for Palantir’s valuation to float back down to something reasonable in which case it might make for a compelling way to invest in an enterprise AI solution that’s growing like mad.