Data status: August 23, 2026. Palantir has become one of the clearest examples of a difficult investing problem: the business can be exceptional while the stock still demands extraordinary future results. Q2 2026 was not merely strong. Revenue rose 93% year over year to $1.935 billion, U.S. commercial revenue increased 149% to $764 million, and U.S. government revenue grew 90% to $809 million. GAAP operating income reached $912 million, a 47% margin.
Those numbers settle an old debate. Palantir is no longer a company that asks investors to believe software economics will eventually appear. They are already visible. The harder question for Palantir stock is whether today’s valuation leaves enough room for the inevitable slowing of growth.
Q2 2026 changed the scale of the discussion
- Revenue: $1.935 billion, up 93% year over year.
- U.S. revenue: $1.573 billion, up 115%.
- U.S. commercial revenue: $764 million, up 149%.
- U.S. government revenue: $809 million, up 90%.
- GAAP operating income: $912 million, a 47% margin.
- Adjusted operating income: $1.194 billion, a 62% margin.
- Deals of at least $1 million: 220.
- Total contract value: $3.373 billion.
- U.S. commercial remaining deal value: $6.238 billion, up 124%.
The important part is the combination. Commercial growth is no longer an experiment running beside a government business. At the same time, government demand is accelerating rather than fading. Palantir is expanding in two customer groups with very different budgets, procurement cycles and operating constraints.
AIP is valuable because enterprise AI is mostly an integration problem
The easiest way to misunderstand Palantir is to call it an AI company and stop there. Large organizations already have access to capable models. Their problem is turning models into reliable actions inside real institutions. Data sits in incompatible systems. Permissions matter. Compliance matters. Every department has different definitions of the same object. A language model alone does not solve that.
Palantir’s advantage is the layer between raw data and operational decisions. Its ontology maps an organization’s objects, relationships, rules and permissions. AIP then connects models to that structure. The result can be a workflow in manufacturing, logistics, healthcare or defense rather than an impressive chatbot with no authority to act.
That is why I see the moat as integration, deployment and trust rather than ownership of the best foundation model. Models will change quickly. The difficult work of making an institution legible to software is much slower to replace.
The bootcamp model solved an old sales problem
Historically, one criticism of Palantir was that deployments looked too bespoke. If every customer required months of custom work, the business would behave more like high-end consulting than scalable software. AIP bootcamps changed the sales motion by compressing the path from demonstration to production.
A customer can bring one painful workflow, connect a useful slice of data and build something tangible in days. That matters because enterprise software purchases often die in committees before users see value. A fast proof of value shortens the distance between curiosity and budget.
The deal count supports the idea that the funnel is broadening. Palantir closed 220 transactions worth at least $1 million in Q2, including 73 worth at least $10 million. I would still separate contract announcements from revenue. The quality signal is expansion after the first workload goes live.
Why U.S. commercial is the decisive segment
U.S. commercial revenue of $764 million grew 149%, while U.S. commercial total contract value reached $2.132 billion, up 153%. Remaining deal value increased to $6.238 billion. That tells me the commercial pipeline is not merely keeping pace with recognized revenue.
Commercial software also has a different compounding mechanism from government contracting. A company can begin with one use case and expand the platform across business units, applications and datasets. Once many workflows depend on the same ontology, the switching cost becomes organizational. Replacing the software means redesigning how the institution understands itself.
The government business is still strategically important
U.S. government revenue grew 90%. I do not treat that as a legacy segment that will eventually become irrelevant. Defense and intelligence customers impose unusually strict requirements around access control, auditability, security and reliability. Product capabilities hardened under those constraints can transfer into regulated commercial markets.
The government business also creates risk. Procurement is political. Budgets change. Programs can be delayed or restructured. A valuation that assumes every major public-sector contract expands smoothly would be too optimistic.
Margins prove the model, but they reduce future easy upside
GAAP operating margin of 47% is remarkable for a company growing at this rate. Adjusted operating margin of 62% is even higher, although I prefer GAAP economics when thinking about valuation. The result confirms that Palantir can scale revenue much faster than costs.
There is a hidden consequence. Once margins are already exceptional, future earnings growth has to come increasingly from revenue. Investors cannot assume both near-100% revenue growth and endless margin expansion. The easier part of operating leverage is already showing up.
The moat has three layers
Ontology. The software creates a structured representation of how an organization operates. The more applications attach to that layer, the more painful replacement becomes.
Deployment speed. Bootcamps are not just a sales tactic. They force the product to prove itself quickly. In enterprise software, the ability to move from prototype to production can matter more than a feature checklist.
Trust. Sensitive AI workloads require permissions, audit trails and control over proprietary information. Palantir has spent years building for environments where those issues are central rather than optional.
Competition is wider than the model layer
Palantir does not need to train the world’s best frontier model. It can integrate models from multiple providers and focus on the operational layer above them. That is strategically attractive because model quality may commoditize faster than enterprise integration.
The challenge is that cloud and data-platform vendors want the same layer. Microsoft can combine Azure, Fabric, Copilot and enterprise identity. Amazon can connect AWS infrastructure, Bedrock and large customer data estates. Databricks, Snowflake and consulting firms can push into workflow orchestration. Palantir’s defense is not exclusivity. It is being better at getting complex organizations into production.
Valuation is the real argument
At a share price around the levels seen in August 2026, the market is not valuing Palantir as an emerging software vendor. It is valuing the company as a future category leader. That changes the burden of proof. A great quarter is no longer enough; the company must keep producing uncommon results for years.
| Scenario | 2029 revenue | FCF margin | Illustrative FCF |
|---|---|---|---|
| Bear | $15bn | 35% | $5.3bn |
| Base | $22bn | 42% | $9.2bn |
| Bull | $30bn | 45% | $13.5bn |
The table is not a price target. It is a way to expose the assumptions. The uncomfortable feature of the bear case is that it still describes a very successful company. Palantir can become much larger and still deliver disappointing shareholder returns if the valuation multiple compresses faster than free cash flow grows.
Stock-based compensation still belongs in the analysis
Stock-based compensation is a real economic cost because it transfers ownership from existing shareholders to employees. I do not object to equity compensation in a technology company, but I track per-share revenue, earnings and free cash flow rather than aggregate figures alone.
If revenue grows rapidly while dilution stays modest, shareholders can still compound very well. If dilution remains structurally high, part of the operating success is absorbed before it reaches each share.
What I would watch next
- U.S. commercial revenue growth and customer expansion.
- Remaining deal value versus recognized revenue.
- GAAP operating margin and free-cash-flow conversion.
- Growth in the number of large commercial contracts.
- Government program concentration.
- Diluted share count.
- Evidence that customers move from one AIP workflow to many.
What would make me more bullish?
I would become more constructive if commercial growth remains exceptionally high while customer concentration falls, remaining deal value continues to outpace revenue and GAAP cash conversion stays strong. I also want to see repeat expansions from customers that began in bootcamps. A meaningful share-price correction without deterioration in those operating metrics would improve the expected return far more than another record quarter at a higher valuation.
What would break the thesis?
The thesis weakens if commercial growth decelerates abruptly, if large cloud platforms make deployment materially easier without Palantir, or if customers build enough internal capability to reduce dependence on the ontology layer. It also weakens if government growth becomes concentrated in a narrow set of programs or if dilution absorbs too much per-share growth.
My conclusion
Palantir has crossed the line from promise to demonstrated economics. Q2 2026 showed a company growing 93% with a 47% GAAP operating margin and extraordinary U.S. commercial momentum. That is rare.
The stock is difficult for the same reason. The market already recognizes much of the quality. My view is therefore positive on the business and disciplined on the entry price. The company does not have to become merely large. It has to remain unusually important inside customer operations for a long time.
That is the central distinction for investors: admiration for the company is not the same thing as an attractive expected return.
Primary sources
This article is analysis, not investment advice. Scenario figures are illustrative.
What the market is really underwriting
At this stage, Palantir is being valued less on what it is today than on the probability that it becomes a deeply embedded operating layer across government and commercial organizations. That distinction matters. A normal software company can miss a quarter and recover. A company priced for category leadership has much less room for execution noise.
The most important valuation question is therefore duration. If U.S. commercial growth remains exceptional for several years and deployment density keeps increasing inside existing customers, the company can grow into a large part of today’s expectations. If growth normalizes sooner, multiple compression can overwhelm otherwise excellent operating results.
Why customer expansion matters more than logo growth
New customer additions attract attention, but expansion revenue is the more important evidence of product value. A customer that begins with one AIP workflow and later deploys the platform across procurement, manufacturing, finance or security creates stronger economics than a customer that never moves beyond a pilot.
This is also where Palantir’s ontology can become a genuine switching cost. The more an enterprise models its workflows, permissions and business objects inside the platform, the harder migration becomes. That is not lock-in by contract. It is lock-in by operational dependence.
Why valuation discipline matters even more for exceptional businesses
A recurring lesson in my book Aktienanalyse mit KI is that business quality and stock attractiveness must be separated. Palantir is an ideal example. The quality is visible in growth, margins and customer demand. The stock question requires a second layer: what assumptions are already embedded in the price?
That is why I prefer a reverse-valuation approach. Instead of asking what multiple Palantir deserves, I ask what future revenue, margin and free-cash-flow path is required to justify the current enterprise value. If those assumptions look heroic, the stock can remain risky even when the company keeps winning.
How I would frame a long-term position
For a diversified portfolio, Palantir should be treated as a high-duration growth asset rather than a defensive software holding. That means position size should reflect the possibility of large multiple compression during periods of rising bond yields or slowing risk appetite. The operating business may remain strong while the stock declines sharply.
This is closely related to the portfolio framework I discuss in Portfoliomanagement mit KI: conviction should not determine position size by itself. Correlation, valuation sensitivity and drawdown risk matter too.
Related reading on The Kapital
For the broader valuation context, see our analysis of why rising bond yields hurt growth stocks. Investors comparing Palantir with other AI infrastructure beneficiaries may also want to read our Micron AI-memory analysis and our Amazon AWS and AI-capex analysis.
FAQ
Is Palantir still mainly a government contractor?
No. Government remains strategically important, but U.S. commercial growth has become one of the strongest parts of the business and is now central to the valuation debate.
What is the biggest risk to Palantir stock?
The biggest risk is not necessarily a collapse in the business. It is that growth slows from extraordinary to merely strong while the valuation multiple compresses.
What metric matters most after revenue growth?
I would watch commercial remaining deal value, repeat customer expansion and GAAP free-cash-flow conversion. Together they show whether growth is durable and economically valuable.
Can Palantir keep margins this high?
Possibly, but the higher margins already become, the less future upside can come from operating leverage. Long-term earnings growth will increasingly depend on revenue growth.
Would I buy Palantir at any price because the business is strong?
No. A great business can be a poor investment when expectations are too high. Entry valuation still matters.
Scenario analysis: what happens if growth normalizes?
A useful stress test is to assume that Palantir remains an excellent company but no longer grows at extraordinary rates. Imagine revenue growth falls from current levels toward 35% to 40%, then toward 25% over the following years. If margins remain strong, earnings can still compound quickly. The problem is that the valuation multiple may fall at the same time.
This double effect is what makes high-duration growth stocks difficult. Slower growth reduces the future cash-flow trajectory, while a higher required return compresses the multiple investors are willing to pay for those cash flows.
Why customer concentration deserves continued attention
Large government and enterprise contracts can create lumpy revenue recognition. Palantir has broadened its customer base, but investors should still examine how much growth depends on a small number of very large deployments. A wider customer mix makes revenue more resilient and reduces the impact of one procurement delay.
What the bull case requires
The bull case requires Palantir to become a standard operating layer for AI-enabled enterprise workflows. That means strong commercial expansion, continued government relevance, low churn and a growing number of customers using the platform for multiple mission-critical processes.
If those conditions hold, the company can justify an unusually high long-term margin structure and a premium valuation. But the bar is high because the market already recognizes much of this possibility.
What the bear case does not require
A bear case does not require Palantir to fail. It only requires the company to grow slower than the market expects. That is an important distinction for investors who confuse a negative stock view with a negative business view.
A company can double revenue over several years and still produce mediocre stock returns if the starting multiple was extreme.
The role of interest rates
Palantir’s valuation is sensitive to real yields because a large share of its perceived value comes from future growth. If long-term bond yields rise, the present value of those future cash flows falls, and investors often demand lower multiples. This can happen even when quarterly results remain excellent.
Final investor checklist
- Is U.S. commercial growth still materially above the software sector?
- Are large customers expanding into multiple workflows?
- Is dilution staying controlled on a per-share basis?
- Are GAAP margins and free cash flow moving together?
- Is the valuation supported by realistic rather than heroic assumptions?
- Would the position size remain comfortable after a 40% drawdown?
Why the commercial flywheel could still surprise to the upside
The strongest upside case is not simply more customers. It is faster deployment followed by broader expansion inside each customer. If bootcamps reduce sales friction and the ontology makes additional workflows easier to launch, revenue per customer can rise faster than the sales organization itself.
That creates operating leverage on two levels: more revenue per account and lower incremental implementation friction as the product becomes standardized.
International commercial is an underappreciated optionality
Most current attention is focused on U.S. commercial growth. If Palantir can reproduce the same adoption pattern across Europe and Asia, the addressable market expands materially. The challenge is that data sovereignty, procurement rules and local competitors differ by region.
I would therefore treat international commercial growth as upside optionality rather than assume the U.S. playbook transfers perfectly.
Why the stock can remain volatile even with strong fundamentals
When a company carries a premium multiple, relatively small changes in expected long-term growth can create large changes in fair value. That means Palantir can report a good quarter and still fall if the market expected an extraordinary quarter.
Investors should therefore distinguish business volatility from valuation volatility. The latter can be much larger than the former.
How to think about dilution over five years
Stock-based compensation should be evaluated cumulatively. A few percentage points of annual dilution may appear manageable in one year, but over five years it can materially reduce each shareholder’s claim on the business. Per-share revenue and free cash flow are therefore essential metrics.
Why competitive intensity could rise
Cloud vendors, data platforms and consulting firms all want to own the enterprise AI workflow layer. Palantir’s current momentum does not guarantee permanent dominance. The company must keep proving that its deployment speed and operational depth justify choosing it instead of a bundled cloud-native alternative.
Long-term conclusion
Palantir is one of the rare companies where the bull and bear cases can both begin with the statement that the business is excellent. The disagreement is about how much excellence is already priced in, how long extraordinary growth can persist and how much per-share value survives dilution.
That is why I would treat the stock as a high-quality but valuation-sensitive compounder rather than a simple momentum bet.
