If you thought AI's biggest regulatory headache was the EU AI Act or a handful of angry senators grilling tech CEOs about chatbots, think again. The Trump administration is reportedly gearing up to sit down with major AI companies as early as next week to discuss something far more direct: a government profit-sharing arrangement tied to AI revenues.

Let that sink in for a second. Not a tax. Not a licensing fee. A profit share. The federal government wants a seat at the table — not just as a regulator, but as a stakeholder.

What We Actually Know

According to reporting from Politico, the Trump administration is moving quickly to open dialogue with AI firms about a framework where the government takes a slice of profits generated by AI systems — presumably those built on publicly funded research, federal data, or infrastructure. The meetings could happen within days.

The details are sparse, which is either because this is still being figured out in real time, or because the people involved don't want the specifics leaking before the handshakes happen. Probably both.

Why This Isn't as Crazy as It Sounds

Before you dismiss this as political theater, consider the actual argument: a massive chunk of the foundational research underpinning modern AI — transformer architectures, RLHF techniques, large-scale compute subsidies — has roots in federally funded academic work and DARPA-adjacent programs. The internet itself was a government project. GPS is a government project. The idea that private companies should be able to monetize infrastructure built on public investment without any reciprocal arrangement isn't some radical left-wing talking point. It's a pretty standard venture capital concept applied at national scale.

That said, the execution risks here are enormous. And the devil, as always, is buried somewhere in the implementation details nobody has written yet.

The Part They're Not Telling You

Here's what the press releases will conveniently skip when this rolls out:

• Defining "AI profit" is a nightmare. Is it revenue from an AI-powered product? Margin attributable specifically to the model? What about AI features bundled into existing software? Every major tech company has lawyers who eat definitional ambiguity for breakfast.
• Enforcement is another beast entirely. The IRS already struggles to track standard corporate tax structures. Auditing AI profit contributions across hyperscalers, startups, and everything in between requires technical expertise the government currently does not have at scale.
• Chilling effects on startups are real. A profit-share regime that's manageable for Microsoft or Google could be existential for a Series B company burning cash on GPU clusters. If the threshold isn't calibrated carefully, you're taxing the ecosystem's growth layer hardest.
• Geopolitical optics matter. China is not slowing down. Any friction added to US AI development — even well-intentioned friction — gets filed under "things that help Beijing" in certain strategic circles.

What AI Companies Are Thinking Right Now

Behind closed doors? A mix of resigned pragmatism and quiet alarm. The big players — your OpenAIs, your Anthropics, your Metas — have enough lobbying infrastructure to shape whatever framework emerges. They'll negotiate, they'll get carve-outs, and they'll ultimately absorb or redirect whatever costs land on them.

The smaller shops are the ones sweating. If you're a founder running lean on inference costs with a product that just hit profitability, the last thing you need is a new federal claim on your margin before you've even hit Series C.

The Bigger Picture

Whatever you think of the politics, this signals something important: AI has moved from "emerging technology we should encourage" to "mature enough to generate revenue we want a piece of." That's actually a milestone worth noting — even if the mechanism being proposed is rough around the edges.

Governments worldwide are wrestling with how to capture value from AI without killing the goose. The UK is doing it through procurement. The EU is doing it through compliance costs. The Trump administration, apparently, is going straight for the P&L.

Whether that's bold pragmatism or a policy framework held together with duct tape and optimism remains to be seen. The meetings haven't happened yet. The details don't exist yet. And in Washington, a lot can change between "as soon as next week" and an actual signed agreement.

Watch this space — and maybe also watch your cap table.