August 22, 2025 - 3:00pm
If you think that a company should be valued at 115 times its revenues — in other words, it would take 115 years of unbroken sales for the company to buy itself back — then you may still be interested in purchasing some shares in Palantir.
In 2005, Google’s price-to-sales ratio peaked at 22 to 1. Today, Palantir’s stock sits at roughly four times that ratio. In recent days, this extraordinary spike has at last begun to wobble, with stock falling by 18%. But even as sanity starts to return, Palantir’s staggering rise tells us much about how the market is pricing AI and the new wave of digital services.
Founded in 2003 by tech investor Peter Thiel alongside CEO Alex Karp, Palantir did AI before AI, collating and structuring data for the US government. But it was only when the AI boom hit a decade and a half later that it was able to brand its skills much more effectively. Today, it has a $10 billion contract to manage the IT and procurement systems of the US Army, not to mention a £330 million stake in the NHS’s Federated Data Platform. These partnerships have appealed to investors. But the secret sauce that has catalysed the company’s limitless growth has been breaking through into corporate America.
Palantir has pioneered a novel “bootcamp” model of IT problem-solving. Senior executives are sent into Palantir and paired with the company’s engineers. Instead of waiting six months for a McKinsey slide deck, executives spend five days inside Palantir. They bring in a real problem on Monday, and leave on Friday with live software, built on real data. The concept has lit up corporate America — so that when the company suddenly doubled revenues to hit its first billion-dollar quarter this month, the markets snapped to attention.
After all, Palantir has real, steady revenue — rarer than one might think in the NASDAQ. The consumer end of the AI boom, like so many other historic booms, has so far been driven by expected returns: by the promise of jam tomorrow. And as with all booms, nine in 10 of the runners and riders today will either be merged or disintegrated in half a decade. The bet is always that one of them, an OpenAI or an ElevenLabs, will “win” their space and start posting mega-profits. For now, most are a flutter.
But Palantir represents the opposite end of the pipeline. The closest parallel is with IBM and its obsession with “mainframes” in the mid-Eighties. While John Q. Public was playing Pac-Man on his 386 IBM computer, in a Morgan Stanley backroom the men in blue suits were winding the magnetic tape that would replace an entire city block of archives. It was those service contracts that kept the company rolling in money for at least a decade, even as its stake in the PC market was eaten away.
Palantir’s product is effectively middleware at scale, but its revenues do not stack in “tokens”, each worth fractions of a penny — the fashionable currency of the API world. They stack into ongoing contracts with the biggest entities in the world, individually worth millions or even billions of dollars.
The socially transformative power of AI is obvious. But as the market has slowly begun to get real about the revenue prospects, Palantir has felt like a safe anchor point. The irony might be that, while the company is due a big correction, it may end up as the smallest correction of all.
Join the discussion
Join like minded readers that support our journalism by becoming a paid subscriber
To join the discussion in the comments, become a paid subscriber.
Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.
Subscribe