Palantir After the Drawdown: What the Operating Leverage Actually Means

Palantir operates three core platforms. Gotham serves government and defense customers, integrating intelligence data, sensor feeds, and operational information into a unified picture. It is the platform the U.S. military and allied intelligence agencies have relied on for counterterrorism and battlefield analysis for over a decade. It is not easily replaced.

Foundry is the commercial equivalent: a data integration and workflow platform for large enterprises. Companies use it to connect fragmented data systems and build operational processes on top of them. Foundry wins in complex, high-stakes environments where off-the-shelf software does not cut it. That selectivity is both the strength and the limitation.

AIP, launched in 2023, is the artificial intelligence layer that sits across both platforms. It allows operators to deploy large language models against real organizational data with the security and audit controls that enterprises and governments actually require. This is not a chatbot product. It is infrastructure for organizations that cannot run AI on public cloud endpoints.

The commercial business has been the laggard historically. In 2021, government revenue dominated. By 2025, commercial growth has accelerated enough to meaningfully shift the revenue mix, and AIP is the primary catalyst. The March 31 announcement of a new Program of Record with the Pentagon reinforces that the government business is not standing still either.

Revenue grew from $1.5 billion in 2021 to $4.5 billion in 2025, a three-year compound annual growth rate of approximately 31%. The pace accelerated in 2025, when revenue grew 55% versus $2.9 billion in 2024. That is not a gradual ramp. Something structural changed.

The 2025 acceleration is primarily AIP-driven. The product launched with a bootcamp sales model in 2023 that was unusual but effective: prospects came in, built working prototypes in days, and signed contracts before leaving. The close rate on bootcamps was high. The deal sizes were large. The cohort of AIP customers that signed in 2023 and 2024 contributed expanding revenues as they deepened usage in 2025.

Gross margin improved from 78% in 2021 to 82.4% in 2025. This matters because it means the revenue growth is not being purchased through lower-margin deals. Palantir has pricing power. The customers who need what Palantir builds do not have credible alternatives at the same level of capability, and the pricing reflects that.

EPS nearly doubled from $0.37 in 2024 to $0.74 in 2025 on an annual basis. The quarterly progression tells the story clearly: $0.09, $0.06, $0.14, $0.13, $0.16, $0.21, $0.24 across seven consecutive quarters, with beats of 12% to 50% in six of those seven periods. The business is executing consistently above expectations.

From 2021 to 2025, operating income went from negative $0.4 billion to positive $1.4 billion. That is a $1.8 billion swing on $3.0 billion of incremental revenue, implying roughly 60 cents of operating income per dollar of new revenue. Most software businesses at this stage of maturity would be proud of 30 cents.

The math behind this is structural. Palantir's core platforms are built. The marginal cost of signing a new customer onto Foundry or AIP is a small fraction of the annual contract value. The heavy engineering investment happened years ago. What is left is largely deployment, customization, and support, which are labor-intensive but not nearly as capital-intensive as building the platform was.

Operating expenses have grown materially in absolute terms as headcount expanded to support AIP's commercial rollout, but they have grown more slowly than revenue. Sales and marketing costs, which were disproportionately high relative to revenue through 2022, have become more efficient as the bootcamp model compressed the sales cycle and the pipeline has become more inbound.

The 2025 operating margin of 31.6% is not a ceiling. If revenue continues compounding at 25% to 30% and fixed cost growth continues at 10% to 15%, operating margins in the 40% to 45% range are reachable within three to four years. The business has the cost structure to get there. The question is whether revenue growth holds.

Free cash flow grew from $0.3 billion in 2021 to $2.1 billion in 2025. That is a sevenfold increase in four years. For a company at this stage of growth, that FCF conversion rate is excellent. It reflects the capital-light model: Palantir spends almost nothing on capital expenditures.

The complication is stock-based compensation. SBC has run between $0.5 billion and $0.8 billion per year since 2021. In 2025, SBC was $0.7 billion against FCF of $2.1 billion, meaning roughly 33% of reported free cash flow was funded by shareholder dilution rather than economic value creation. If you subtract SBC from FCF to get a truer picture of cash earnings, 2025 FCF adjusts down to approximately $1.4 billion.

The dilution record confirms the cost. Share count grew from 1.76 billion in 2020 to 2.53 billion by end of 2024, an increase of 44% over four years. Shareholders who held through the company's growth have been systematically diluted. Buybacks have been minimal, totaling roughly $0.1 billion in both 2024 and 2025, which does not come close to offsetting annual SBC issuance.

This is improving. SBC as a percentage of revenue fell from 53% in 2021 to 16% in 2025 as the revenue base scaled. But at current share prices, the absolute dollar dilution is meaningful enough that any valuation model should use SBC-adjusted FCF rather than reported FCF. The distinction matters at a $329 billion market cap.

On March 31, Palantir announced a new Program of Record with the Pentagon, a designation that carries specific significance in defense contracting. Programs of Record are formally budgeted, multi-year commitments. They are not experimental engagements or pilot programs. They are embedded in the defense budget process and difficult to cancel outside of major policy shifts.

This matters because the most persistent risk cited against Palantir's government segment has been concentration risk combined with budget uncertainty. A Program of Record changes the stability profile of that revenue. Once embedded in a Program of Record, Palantir's platform becomes infrastructure that the military cannot easily unwind without operational disruption.

The government segment already showed acceleration in 2025, contributing meaningfully to the overall revenue jump from $2.9 billion to $4.5 billion. The commercial segment has been the growth narrative, but the government segment is proving it can grow at the same time rather than trading off against it.

The question facing defense-oriented revenue is the DOGE-related scrutiny on government technology spending. If discretionary defense technology budgets face compression, Palantir's newer government contracts could see slower expansion. A Program of Record designation provides insulation. The core Gotham mission-critical work is the most insulated of all.

Palantir's competitive position rests on three interlocking advantages. First, the classification ceiling: Palantir is cleared to operate in environments that hyperscalers and most enterprise software vendors are not. That accreditation is expensive and slow to replicate. It creates a structural barrier against cloud-native competitors for the government segment.

Second, data architecture depth: Palantir's platforms integrate data at a level of complexity that most enterprise tools cannot match. Foundry's ontology layer, which maps relationships between entities across an organization's data, is not a feature that competitors can bolt on. It requires rethinking the data architecture from the foundation.

Third, switching costs: once Palantir's platforms are embedded in operational workflows, particularly in government or industrial settings where the cost of error is severe, switching becomes an organizational risk that most customers prefer to avoid. The enterprise sales cycle is long. The exit cycle is longer.

The limitation is that these advantages apply most cleanly to complex, high-stakes, high-budget customers. The middle market has cheaper options. Palantir has never been a volume business and shows no signs of becoming one. If AIP's bootcamp model has opened a broader commercial market, the next two years will determine whether it scales into mid-market or remains a large-enterprise play.

At $146 per share and a $329 billion market cap, Palantir trades at 73x trailing price-to-sales, 218x trailing earnings, and an EV/EBITDA of 223x. These are not rounding errors. This is a stock priced for an outcome several multiples larger than the current business.

To justify the current valuation using a free cash flow framework, you need Palantir to compound SBC-adjusted FCF at approximately 30% annually for ten years, then apply a terminal multiple of 30x. On those assumptions, working backward from a required discount rate of 10%, the current price is defensible but not comfortable. Change the revenue growth assumption to 25% or the terminal multiple to 20x, and the stock is materially overvalued.

The analyst community reflects this uncertainty. Fifteen analysts rate PLTR a hold. Four rate it a strong sell. Only four rate it a strong buy or buy. The consensus price target is $186.60, implying roughly 27% upside from current levels, but the distribution of that consensus is wide, and the strong sell camp is credible rather than reflexively bearish.

The sentiment trend offers a different angle. Normalized sentiment scores fell from above 0.90 in mid-March to 0.33 to 0.48 in early March, coinciding with the broader technology selloff and concerns about government spending cuts. Sentiment recovered sharply by late March, back above 0.90. The underlying financials did not change during this period. What changed was macro sentiment. That volatility is inherent in a stock where the valuation leaves no margin for error on the narrative.