Palantir Under Surveillance
A deep dive into the areas of $PLTR where most investors don't look.
This post is a practice case study building on the prompt analysis techniques taught in Part 1 and Part 2 of our forensic series “AI vs the 10-K”. Download the PDF output report at the bottom so you can visualize the analysis you can perform on any stock in minutes.
Important Note: The Inferential Investor does not provide financial advice. Any discussions on stock and market views are made from an educational standpoint and to demonstrate techniques and capabilities of AI, are indicative and may be subject to change. No recommendation, either direct or implied is made with regards to the suitability of any stock mentioned for investment. It is recommended to refer to our full disclaimer here and always do your own research, but I welcome you to follow this substack as another input into a well rounded investment process.
Any stock trading at 100x Price/Sales demands hard questions…
Palantir ($PLTR) has been a monster stock built on the AI adoption wave with successive revenue beats, guidance upgrades and investor excitement over an accelerating growth rate. This heady combination has seen the stock rally 400% in the last twelve months to trade at forward multiples of ~100x Price/Sales and ~250x Price/Earnings. At valuations like that, there cannot afford to be red flags hidden within the accounting detail, so it is a worthwhile exercise to conduct a forensic examination of PLTR’s latest annual report to see what surfaces.
This exercise was undertaken in the spirit of sound and skeptical inquiry. We applied our Forensic Security Analysis prompt - a tool designed to bring an investor closer to the operational core of a business and go deeper than the superficiality of reported revenue and earnings growth. Palantir’s financials were subjected to rigorous examination, including tests of accrual intensity, earnings quality, the consistency of free cash flow, and the appropriateness of revenue recognition practices. We scrutinized capitalization policies, working capital behavior, leverage metrics, and covenant conditions with each a diagnostic intended to reveal where the numbers may diverge from economic reality.
The final report on Palantir, generated by II’s prompt techniques, has been attached for download at the end of this post and makes for a fascinating read. The model output demonstrates the capabilities of Inferential Investor’s prompt sets and the insights they generate. These tests take less than five minutes to run despite processing over one hundred pages of source material - an exercise in efficiency and discipline that any serious investor should find worthwhile.
The Forensic Tests
A cornerstone of prudent investing is the examination of a company’s financial records. While a company's quarterly presentation may capture the headlines, it is the underlying figures that reveal the true quality of the enterprise. In a market often driven by fleeting narratives, the careful investor should look beyond the headlines and engage in a disciplined, forensic analysis of the financial statements, particularly when it is as easy to complete as AI now makes it. This is especially true for a company like Palantir Technologies, whose work in both government and commercial sectors is often as complex as its financial reporting and whose extreme valuation, while a mark of success, is also its largest risk.
An initial review of Palantir’s 2024 financial performance reveals a powerful, yet nuanced, story. The Forensic report is pleasingly, almost surprisingly, clean of red flags. Palantir demonstrates a conservatism in its reporting choices that is refreshing in the entrepreneurial and highly competitive AI world. Earnings and cashflows are of high quality with accruals, based on the tests conducted by the model, not presenting any major concerns. Research and development is fully expensed, a conservative management choice, working capital is well managed and the company operates with significant liquidity and is debt free. There are material off-balance sheet contractual purchase commitments of $2 billion but these relate to long dated cloud hosting services that are critical for the company operations and matched by revenue contracts with high quality counterparties. It is in this respect where Palantir’s shadow leverage resides and one reason the company likely avoids significant financial debt.
The most significant finding for the diligent investor is the marked divergence between the company’s reported GAAP net income and its cash flow from operations (CFO). In 2024, the company recorded net income of approximately $467.9 million, a commendable figure that marked a significant turnaround from prior losses. However, a deeper look at the cash flow statement shows that the true cash-generating power of the business was far greater, with cash flow from operations reaching an impressive $1.15 billion.
This is not a sign of low-quality earnings, as is often the case when this gap exists. Instead, it is a direct consequence of a deliberate business decision: the heavy reliance on non-cash stock-based compensation (SBC). With an SBC expense of $691.6 million in 2024 alone, a substantial portion of the company's profitability is tied not to cash outflows but to equity grants. For the intelligent investor, this distinction is critical. It suggests that while GAAP earnings may be a useful measure, cash-based metrics provide a more complete picture of the company’s operational health and, importantly, its ability to fund its own growth without external financing. The business, in its operational core, is highly cash-generative and requires minimal capital expenditure to grow.
The analysis also points to a successful strategic pivot. Palantir has long been known for its work in the government sector, characterized by large, complex contracts. However, the data now shows a new engine of growth: the U.S. commercial market. In 2024, U.S. commercial revenue expanded by an impressive 54% year-over-year which accelerated to over 90% by the second quarter of this financial year, and, just as importantly, the contribution margin of this segment grew from 52% to 60%. Accelerating growth and simultaneous margin expansion is a powerful combination as the stock price already shows. This is evidence of genuine operational leverage, suggesting that the company is finding a more scalable and efficient way to deliver its services to a wider range of customers. For the observer of sound business principles, this is a clear indication that the business model is maturing and becoming more robust.
The Benefits and Costs of Stock Based Compensation (“SBC”)
However, a closer look at the capital structure reveals a critical challenge. The very stock-based compensation that inflates cash earnings (when compared with paying this out as cash compensation to employees) also results in significant dilution for shareholders. From 2022 to 2024, the company's weighted-average diluted share count swelled from approximately 2.06 billion to 2.45 billion. This constant issuance of new shares acts as a headwind to per-share value creation.
While the company has initiated a share repurchase program to address this, its rate of deployment remains a key question and its magnitude is simply inadequate to stop dilution. The primary reliance on stock as a form of compensation (at a level that equates to 16% of company revenue, 30% of income and somewhere between 30% and 40% of the estimated and undisclosed cash employee compensation component) is an explicit trade-off: it is highly effective at attracting and retaining talent in a rising market and removes a large cash expense to assist earnings, but becomes a significant risk factor in a declining one.
Let’s clear up some common misconceptions about stock-based compensation (SBC) before elucidating that last point about the risks behind SBC. The annual expense companies report (tucked “below the line”) is a conservative estimate of the value transferred - at least from a shareholder’s perspective. That $692 million line item for Palantir, isn’t simply the number of shares granted multiplied by the share price at the grant date. And that’s exactly why management loves SBC while shareholders loathe it.
For employees, SBC feels far more valuable than the modest hit to the income statement. When they see their company growing, they mentally calculate vested shares multiplied by a future share price - a total value often 10 times higher than the pro-rata expense booked for them that year. From their vantage point, it’s a terrific deal: give an employee “$1,” and in a rising market they perceive it as $10. The annual accounting expense, meanwhile, is just an option-style estimate that assumes vesting risk and discounts the upside.
Shareholders, though, live in a different reality. Management often encourages them to ignore SBC in “adjusted” earnings, arguing it isn’t a cash cost. But behind that $692 million is the looming issuance of a huge block of stock that dilutes existing investors. At the end of FY24, Palantir had 250 million shares - roughly another 10% of its common stock - waiting in the wings under SBC programs. On a $400 billion market cap, that dilution equates to about $40 billion in value quietly siphoned from stockholders. The multiplier from a few years of SBC expense to that cumulative $40bn is immense! It’s less a line-item expense than a magician’s trick: watch the earnings hand while the dilution hand rearranges ownership.
Outside simple ongoing annual dilution, there’s another risk. If the company's stock, which trades at an extreme multiple currently, were to suffer a material decline, the value of that employee compensation would be directly impacted. To maintain the competitiveness of their compensation packages and to retain talent who’ve just seen the wealth they had banked in their mind evaporate (options become worthless), the company would be compelled to issue an even greater number of shares at the new, lower price to its employees, possibly with a “catch-up” or “make good” component to irreplaceable talent. This escalates the dilution of existing shareholders at the exact moment they are being hurt the most by the decline in value and exacerbates downside risk. The cost of retaining talent in such a scenario is effectively borne by shareholders through the increased dilution of their ownership stake. An investor's returns are earned on a per-share basis, and a company that continually issues new shares, particularly to offset a falling price, must be watched with vigilance - in normal times, it reduces upside and in a downturn it exacerbates downside - a really bad deal for shareholders.
Hidden Overvaluation
This identification of stock based compensation as an issue can be overlooked by many investors who have become accustomed to seeing this dynamic across technology stocks. However the way it should be considered is as certain future dilution of the share count that needs to be taken into account in up-front valuations.
After all, if you are using cash based earnings (normalizing the SBC expense) as the basis of valuation, then it follows that the investor must take ongoing dilution patterns into account on the equity side of the valuation equation. By this, I don’t mean just using the “fully diluted share count” reported in the financial statements. If it is a company’s policy to grant new options each year as a normal part of compensation, then investors should be taking that predictable growth in fully diluted shares into account, as those shares will exist when it comes time to sell.
Just like cash employee compensation is considered a recurring expense and taken into account in recurring earnings for valuation purposes, stock based compensation is a recurring dilution and as a known factor, should be discounted into valuations.
If an investor was considering buying PLTR today with a five year horizon, they need to be extending the rate of share count increase that has been growing at up to 200m shares or ~10% of the total share count p.a. The buyback will reduce this growth but not stop it. If, in a purely hypothetical extrapolation of past share count trends, there were to be 50% more shares on issue over the next 5 years for PLTR, a prudent 5-year investor would be contemplating that the stock is not actually on 100x Price / Sales or 246x Price/Earnings that their screen tells them, its actually on 150x P/S and 370x P/E!. At these levels, numbers seem to become slightly meaningless. Let me put that in simple terms - Palantir, if it continued to issue employee stock as it has in the past, could be considered to be valued (on an exit equivalent basis) at nearly 400 years of profits!
At these valuation levels we should expect that every single vested share that employees have access to, will be sold the moment they have a chance. The downside risk to their primary source of wealth is just too great.
Closing the Forensic Microscope on Palantir
The diligent study of the financial statements reveals a company of contradictions: immensely profitable on a cash basis and showing clean and conservative policies on financial reporting, yet addicted to persistent dilution. Successfully pivoting to new markets, but still facing scrutiny over its governance and related-party transactions, such as a $7.7 million payment to its CEO for personal aircraft use.
For the thoughtful investor, the path forward is one of continued observation. At these valuations, where the biggest risk to the company is its own share price, this is not a time for reckless speculation, but for a sober evaluation of the facts, grounded in the belief that in the long run, the underlying economics of the business will determine its ultimate worth.
I encourage you to review the 33 page downloadable report for II’s forensic analysis prompt, applied to Palantir. This demonstrates the detailed insights these structured professional analysis techniques can achieve for any investor.
As always,
Inference never stops. Neither should you.
Andy West
The Inferential Investor