Palantir Built a Company Around a Manifesto. The Manifesto Is Also the Risk.

In April 2026 Palantir did something most enterprise-software companies would treat as career-ending: it posted a 22-point manifesto declaring that some cultures are 'middling, and worse, regressive and harmful,' that Silicon Valley owes the United States a 'moral debt,' and that its engineers have an 'affirmative obligation' to help defend the nation. It racked up 32 million views.⁵ No company sells more software by insulting half the planet's worldview. Palantir does not appear to care - and that indifference is not a bug in its strategy. It is the strategy.

The easy reading is that this was a provocation, a CEO with a book to sell turning the volume up. That reading is wrong on the facts. The posture isn't new and it isn't theater. It was written into Palantir's founding documents, and it has been quietly screening who the company hires, who it sells to, and which contracts it chases for two decades.

The culture is in the SEC filings, not the swag closet

Most companies treat culture as a recruiting page - free lunch, ping-pong, a values poster nobody reads. Palantir treats it as load-bearing engineering infrastructure. When it filed to go public in 2020, it did not bury its ideology in a charity footnote. It put it in the prospectus, the most lawyered document a company ever produces, where every word carries liability.

Read that twice. Not a result of the software. A precondition for it. The same filing took a public side - against consumer internet companies that sell user data, and explicitly for Western democracies and open societies.¹ Its own responsible-business page says the company was 'founded to help Western democracies and open societies harness data.'⁶ So when the 2026 manifesto landed, anyone calling it a surprise simply hadn't read the IPO paperwork. The doctrine didn't change. The microphone got louder.

Here is the thesis, stated plainly: Palantir's culture doctrine is a genuine operating system, not branding - and the very thing that makes it a moat is the same thing that makes it a permanent liability. It is both the engine and the crack in the hull, and they are inseparable, because they are the same mechanism running in two directions.

How an ideology becomes a moat

The mechanism is self-selection. A company that publicly says 'some cultures are harmful' and 'we are at war with China' is performing a filter. The people who find that repellent never apply. The people who find it electric - engineers who want to point their careers at a mission rather than an ad-targeting funnel - sort themselves into the building. You don't have to manufacture mission-alignment if the front door does it for you.

And the numbers say the filter works. Palantir grew from 313 people in 2010 to 3,936 by the end of 2024.² By FY2025 it cleared roughly $4.5 billion in sales, more than half from government contracts, with revenue per employee near $1 million.⁸ Its CTO can stand on an earnings call and say 'we are at war with China' in the same breath as announcing 32 new deals over $10 million each - and the market rewarded it.⁷ The ideology is not separate from the commercial result. The defense-first, mission-saturated doctrine is precisely what makes a defense-and-intelligence customer trust them with the most sensitive data a government holds. The belief is the sales pitch.

The same filter that builds the moat picks the company's pocket

A filter that screens for belief screens out everything else - including, as it turns out, the law. Palantir's celebrated meritocracy ran heavily on employee referrals: hire the people your best people vouch for. It sounds like pure signal. But a referral network reproduces whoever is already inside it, and in 2016 the U.S. Department of Labor sued Palantir for exactly that. The complaint alleged the hiring process discriminated against Asian applicants for engineering roles, with a referral system that 'routinely eliminated' qualified Asian candidates 'despite being as qualified as white applicants.'³

Palantir settled in April 2017 for $1,659,434 in back wages and other relief, without admitting wrongdoing; the review period ran from January 2010 to June 2011.⁴ Sit with the irony. The mechanism the company sells as proof of pure merit produced a federally documented pattern of the opposite. The 'just hire the best' story has a regulatory finding stapled to it that the sanitized version always leaves out.

The fractures aren't only legal. Palantir's own FY2024 10-K warns that scaling its 'unique sales model' risks compromising the 'cultural and mission-oriented elements' of the company.² In plain English: the thing that made us special might not survive getting big. A culture tuned for a few hundred zealots is hard to stamp onto thousands of new hires, and harder still onto a commercial sales force that joined for a paycheck rather than a crusade. The doctrine that recruits true believers has nothing to offer the people it needs but doesn't inspire.

Maybe the contradictions are the whole point

The fair objection is that this is too tidy - that I'm calling a wildly successful company self-sabotaging while it compounds at nearly 30% a year. If alienating half the market were a real cost, where is it in the revenue? More than half of $4.5 billion comes from government, exactly the buyer the doctrine is built to win.⁸ By that math the manifesto isn't a bug; it's targeting. You don't sell battlefield software to a defense ministry by being studiously neutral about whether the West should win. The polarization is the qualification.

That objection is partly right, and it's why the doctrine is a genuine moat rather than a vanity. But it confuses 'works so far' with 'works without cost.' The government concentration the doctrine produces is itself the risk: a company whose entire identity forecloses China and much of the politically cautious enterprise market has voluntarily capped its own addressable world. The settlement was real money and reputational damage. The 10-K's own warning about losing the culture at scale is the company conceding, in writing, that the engine has a redline. A moat that only holds while you stay narrow and ideological isn't a flaw-free moat. It's a bet that the war never ends - and that you never need the customers who'd rather not pick a side.

Palantir didn't stumble into a culture; it engineered one and put it in its securities filings, where lying is illegal and conviction is expensive. The manifesto everyone treated as a stunt was just the operating system finally said out loud. It built a $4.5 billion company by being the kind of place only certain people will join and only certain governments will trust - and that same exacting filter is what got it sued, what its own 10-K warns may not survive scale, and what walls it off from the half of the market that wants its software without its sermon. The doctrine is the moat. The doctrine is the risk. Palantir's bet is that you can keep both - that the wall keeping the wrong customers out is the same wall keeping the right ones loyal. It's a real bet, and so far it's winning. But you cannot turn the filter off without losing the thing it filters for.