There are growth companies, and then there is Palantir's second quarter. Revenue increased 93% year over year to $1.94 billion. US revenue rose 115% to $1.57 billion, while US commercial revenue jumped 149% to $764 million. The company closed 220 deals worth at least $1 million during the quarter; 73 were worth $10 million or more.
Those figures make it increasingly difficult to argue that enterprise demand for Palantir's artificial-intelligence software is merely another temporary AI spending cycle. They also expose the next question in the investment case.
America is doing most of the work
Palantir's US government business produced $809 million of revenue in the quarter, up 90%. Its US commercial business produced another $764 million. Together, those two businesses account for the overwhelming majority of Palantir's quarterly revenue.
That is impressive because the US market is enormous and both sides of the company are accelerating at the same time. It is also concentration — the same structural feature now visible at index level, where a handful of names are doing most of the work.
The more Palantir's valuation assumes years of exceptional growth, the more important it becomes that the company eventually proves its model can travel beyond the market currently doing most of the heavy lifting. International expansion therefore deserves to be watched almost as closely as US commercial growth.
The contract pipeline remains formidable
There is little evidence yet of American demand cooling. US commercial remaining deal value reached $6.24 billion, up 124% year over year. Total contract value in US commercial hit a quarterly record of $2.13 billion, an increase of 153%. Adjusted free cash flow was $1.22 billion, giving the company a 63% adjusted free-cash-flow margin for the period.
That combination is unusual. Investors normally tolerate enormous valuations in young software companies because current profitability is sacrificed for future scale. Palantir increasingly offers both growth and cash generation, at a moment when the hardware layer beneath AI software is itself supply-constrained. That makes the bull case straightforward to understand.
The risk has moved
Palantir's investment argument used to revolve around whether commercial customers would adopt the platform at all. That question is becoming outdated. Businesses are adopting it, revenue is scaling and large contracts are multiplying.
The harder debate is now about duration. How long can growth remain exceptional? How much of the opportunity can come from outside the US? How dependent is the company on continued expansion in government and defence spending? And how much future perfection is already reflected in investor expectations?
Those are better problems for a company to have. They are still problems investors need to price. For Palantir, the next phase of the story may not be proving that AI demand is real. It will be proving that one of the fastest-growing large software businesses in the market can eventually become as geographically broad as its valuation implies.