Two former Palantir employees reconnected by phone earlier this year. One of them opened the conversation with, "Hey, are you tracking Palunteers's descent into fascism?" That was the greeting. Not, "Hey, how's it going? How have you been? What are you working on now?" Just, "Are you tracking the descent into fascism?" When two people who built your system start using that kind of language, something fundamental has shifted. I'm El. I have a PhD in computer science and I analyze AI developments to understand what's actually happening beneath all of this hype. What is unfolding inside Palunteer right now tells us something super important about what happens when surveillance infrastructure meets ideological certainty.

Just for brief context, Palunteer builds data integration platforms. That's the boring way to say it. The more accurate way is they build software that takes information scattered across dozens of incompatible databases and makes it searchable, mappable, analyzable in real time. The company was founded in 2003, named after the allseeing ors from the Lord of the Rings. Technically, they're from the Sylmerelion made by Feyenor, my hero, but let's cut the daydreaming. Basically, the ones that let you see across vast distances, but also in the books kind of slowly corrupt anyone who uses them. One could argue the corruption was done by Sauron, but that's pretty irrelevant for this video. It's basically quite subtle branding by the company. Their customer list is the CIA, ICE, the Department of Homeland Security, the UK's National Health Service, the US military, that kind of people. They've held government contracts worth billions, positioning themselves as essential infrastructure for national security and public sector operations. And for two decades, Palunteer employees could rationalize the external criticism. Privacy advocates push back. Civil liberties group questioned the work. But internally, the framing basically held. We're the ones making sure surveillance power doesn't get abused. We're building safeguards. We're the good guys in a complicated situation. Well, that framing just flipped.

And here's what changed. In January 2026, ICE agents shot and killed Alex Prey, a 37year-old intensive care nurse in Minneapolis. He was filming law enforcement with his phone, directing traffic around a protest. Bystander video shows that he was unarmed, just holding his phone when he was shot multiple times. I'm sure you all remember that one. Palunteer software powers ICE operations. Specifically, they built something called Elite, a tool that scrapes Medicaid databases and other government records to generate deportation leads. It creates dossier on people, assigns confidence scores to their home addresses, maps potential targets. Nearly 8 million Medicaid patients data were repurposed for immigration enforcement. When Pratty was killed, Palunteer employees started asking questions in internal Slack channels. Basic questions. Were we involved? Are we doing anything to stop or repeat? The company's response was, according to Wired reporting, Palanteer started autodeing conversations in at least one Slack channel after 7 days. Just stop and think about that for just a second. The company that sells data analytics and transparency to governments began scrubbing its own employees questions about whether their own software helped facilitate a killing.

Then in late February, a US air strike hit a girl's elementary school in Iran. 120 school children died. Multiple investigations concluded the strike relied on outdated targeting data. The school had been walled off from a military compound over a decade ago, but the target list was never updated. Precision weapons 10-year-old spreadsheet. Palunteer's platforms are embedded in US military targeting systems. Again, employees asked internally, "Were we involved?" CEO Alex Karp's solution was that he released an hour-long pre-recorded video that according to reporters avoided specifics about product capabilities or ISIS use of Palunteer software. He offered employees one-on-one briefings, but only if they signed NDAs first. You can't ask questions about whether your software killed people unless you legally agree never to tell anyone the answers. What we're watching is the difference between external criticism and internal crisis. External pressure is manageable when your workforce believes in the mission, but when the people actually building the system start openly questioning whether they're enabling harm, that kind of sounds like structural instability.

One former employee described it to Wired this way. I'm quoting the broad story of Palanteer as told to itself and to employees was that coming out of 9/11 we knew there was going to be this big push for safety and we were worried that safety might infringe on civil liberties and now the threat is coming from within. We were supposed to be the ones preventing these abuses and now we're not preventing them. We seem to be enabling them. Meanwhile, Palanteer's leadership doubled down earlier this month. They posted a 22-point manifesto summarizing CEO Karp's book. We have a separate long video analysis on that. Check it out if you like. A direct quote from that is this new dogma glosses over the fact that certain cultures and indeed subcultures have produced wonders. Others have proven middling and worse regressive and harmful. Silicon Valley defense contractors publicly ranking cultures as superior or inferior. Literally the ideological framework every major 20th century atrocity was justified with just dressed up in startup language. Karp's response to employee disscent was if you have a position that does not cost you even to lose an employee it's not a position. He's treating attrition as proof of moral courage instead of what it actually is institutional collapse.

And here's the valuation absurdity. Palanteer currently trades at a price toearnings ratio around 212. The S&P 500 average is 26. The stock is priced eight times higher than the market, assuming the company will maintain current profitability for over two centuries. Investors are literally betting Palunteer will outlive the United States itself while its own engineers are discussing fascism in the breakroom. The company has a $340 billion market cap on $4.5 billion in annual revenue. For comparison, that is a higher valuation than most S&P 500 companies despite generating less revenue than a decent regional grocery chain.

I'm actually developing a course in two tiers for nontechnical professionals where AI is explained simply, not dumbed down, just clarified. If you'd like to take a two minutes to answer some questions which directly shape where the course covers, the link is down below. So, where does all of this go? We've seen this pattern before. Actually, back in 2018, Google employees pushed back against Project Maven AI for military drone footage analysis. Google withdrew from the contract, but Google had diversification options. Consumer products, cloud services, advertising. Palunteer doesn't have that. 60% of their revenue comes from government contracts. The work their employees are questioning is the business model. And in terms of what's going to happen next, either Palunteer loses the talent war, the engineers who can build those systems decide the reputational cost just isn't worth it and leave for enterprise SAS companies, or they become a pure ideological filter, hiring only people who already agree with the mission.

If it's the first, execution risk becomes a real problem for their government customers. You cannot really deliver cutting edge AI systems if your institutional knowledge keeps walking out the door. If it's the second you get technical stagnation, ideological homogeneity, and engineering excellence rarely coexist, you really need people who question your assumptions and you just fire them all. There is a third possibility, and it's the one Palanteer seems to be betting on that they've embedded themselves so deeply into government infrastructure that it doesn't even matter. That they've become too essential to replace regardless of internal turmoil or external criticism.

But here's the thing about building surveillance systems in a democracy, and I mean actual democracies, not the rhetorical version. When your value to the government is primarily ideological alignment with a specific political project, you become a visible target the moment that project loses power. Fascism as a business strategy only works if the fascists stay in power indefinitely. And in the United States, that historically has been a very bad bet. What we're watching is a test case for something genuinely significant. Can a company that builds infrastructure for state power survive when the people building it stop believing in the mission? Can you maintain technical excellence when your engineers open conversations with, hey, are you tracking the descent into fascism? The answer will tell us a lot about where the line actually is between building tools and building weapons and whether that line still matters to the people writing the code.