Palantir (PLTR): the business is extraordinary. The valuation requires a miracle.

Palantir Technologies Inc. (NASDAQ: PLTR) is the most argued-about stock in retail investing right now. One side says it's the AI infrastructure company of the next decade, that the government and commercial contracts are compounding into something enormous, and that the valuation will look cheap in hindsight. The other side says it's a software company priced like a guaranteed monopoly when no such guarantee exists. Both sides have evidence. That's what makes it worth breaking down properly.

What they actually reported — the numbers are real

Q2 2026: $1.935 billion in revenue, up 93% year-over-year. US commercial revenue: $764 million, up 149%. US government revenue: $809 million, up 90%. Adjusted EPS: $0.41, beating the $0.34 consensus. GAAP net income: $1.062 billion, up 225% year-over-year. GAAP operating margin: 47%. Adjusted operating margin: 62%. Free cash flow: $1.22 billion, representing 63% of revenue. Rule of 40 score: 155% — combining 93% revenue growth and 62% adjusted operating margin. That score puts Palantir in a category occupied by almost no other enterprise software company at this scale. Management raised full-year guidance to $8.15 billion to $8.16 billion, implying 82% annual revenue growth. CEO Alex Karp called the quarter "out of this world." He wasn't exaggerating.

Total contract value of $3.37 billion, up 49%. US commercial TCV: $2.13 billion, up 153%. Net dollar retention rate: 157% — meaning existing customers expanded their Palantir spending by 57% over the prior year on average. That's the kind of number that means customers who already use the product are deepening their commitment rather than looking for alternatives. 653 US commercial customers, up from 385 a year ago. 220 deals over $1 million in the quarter. These are not vanity metrics — they're the building blocks of durable revenue.

What AIP actually does and why enterprises are paying for it

The Artificial Intelligence Platform, or AIP, is the product driving the US commercial acceleration. It's not a large language model — Palantir doesn't make foundation models. AIP is an operating layer that sits on top of existing AI models (GPT, Claude, Gemini, Llama) and connects them to a company's actual operational data and workflows in a governed, auditable way. The value proposition: an AI assistant that can access your company's proprietary databases, ERP systems, supply chain data, and operational records without leaking that data to the model provider's training pipeline, without hallucinating information from outside your data, and with a full audit trail of who asked what and what action was taken.

For regulated industries — defense, healthcare, finance, manufacturing — that governance layer is often the difference between a useful AI deployment and a compliance violation. The boot camps Palantir runs, where enterprise teams spend several days building AIP applications with Palantir engineers, have converted into commercial contracts at a rate that management describes as unprecedented. The 149% US commercial growth in a single quarter is a direct output of those boot camp conversions scaling. I'd argue the enterprise AI governance market is larger than most analysts had modeled a year ago — the Q2 numbers suggest that's clearly true.

The bear case — the valuation math is difficult

At $180 per share, Palantir trades at approximately 107 times trailing GAAP earnings and roughly 20 times trailing sales. The optimistic model from TIKR requires approximately 25% annual EPS growth sustained through 2035 just to reach the current price on a forward basis. That's not impossibly fast for Palantir's current trajectory — EPS grew 225% year-over-year in Q2 — but sustaining 25% annually for a decade is a requirement that almost no company in S&P 500 history has met. At some point the growth rate has to moderate as the company gets larger, as competition enters, as government contract cycles fluctuate. The model is not forgiving of any of those things.

HSBC trimmed their target from $205 to $151 and cited "intensifying competition in AI software and the risk that lower barriers to entry could weigh on Palantir's pricing power." That's the bear case in precise terms: if other enterprise AI platforms can do what AIP does at lower cost as model prices fall, the premium Palantir charges compresses. The company's ontology approach — the way AIP structures a company's data for AI access — is designed to be the durable moat. Whether it actually is durable when well-capitalized competitors enter is the open question the current multiple assumes a favorable answer to.

The stock has also declined roughly 31% from its peak at around $265, which tells you the market has already had second thoughts about the most extreme valuations. At $180 it's cheaper than it was, but it's still priced at a level where execution has to remain extraordinary for the math to work.

The government business — the part retail ignores

US government revenue of $809 million, up 90% year-over-year, is the other half of the story. Palantir's relationship with the US defense and intelligence apparatus goes back to 2003 — they've been building war-fighting and intelligence-analysis software for the military and CIA for over twenty years. The US government is not going to switch away from Palantir's Gotham platform the way an enterprise might switch productivity software. The switching cost for classified defense and intelligence infrastructure is essentially infinite. That sticky, recurring government revenue base is the part of the business that actually provides a floor under the revenue trajectory and makes the 82% full-year growth guidance more credible, not less.

The Defense Intelligence Agency contract awarded in Q2, and the expanding Maven Smart System deployment with the US Army, are the kinds of multi-year contracts that don't require quarterly re-selling. Palantir wins a major government contract and it compounds for years. The commercial business gets more attention because the growth rate is higher right now, but the government moat is the more durable structural advantage and it shouldn't be discounted in any valuation framework that takes the long view.

Where both sides are right

Bulls are right that the business is exceptional, the growth rate is real, the customer expansion is accelerating, and the Rule of 40 score of 155% is a level that almost nothing else in enterprise software has reached. If the next decade looks remotely like Q2 2026, the stock could look cheap at current prices in hindsight.

Bears are right that the valuation is pricing in a version of the future where 25 consecutive years of strong execution occur without meaningful competitive disruption, model commoditization eating the margins, or government budget cycles creating revenue gaps. All of those things have happened to enterprise software companies before. None of them are obviously going to happen to Palantir. But the price gives you zero margin of error if any of them do.

The one thing beginners should take away

Palantir is one of the most genuinely exceptional software businesses on earth right now. That's real and documented by the numbers. Buying the stock at 107 times earnings means you're paying for the next decade of that exceptionalism upfront, at full price, with no discount for the uncertainty about whether it continues. Some investments that look overvalued at the time turn out to have been cheap. Most don't. The question isn't whether Palantir is a great company. It clearly is. The question is whether "great company" is the same as "stock worth buying at any price." Those are different questions. The first has a clear answer. The second doesn't.