Schneier Calls For US Nationalization Of AI Giants

The proposal reads as a safety valve: if the market walks away from OpenAI or Anthropic, the US government should be ready to step in and run the thing rather than let a frontier AI lab go the way of a failed regional bank, unmourned and unsupervised. Bruce Schneier and Nathan E Sanders are not proposing this lightly - they are naming the two companies most people would agree sit closest to whatever “frontier” currently means, and they are framing the stakes as a choice between democratic stewardship and letting corporate profit motives steer something that can steer everything else.

One notices that Google and Meta are in the room too - named in the same breath, part of the same industry, building models with comparable or greater capability - and are simply not covered by the proposal. That is not an oversight. It is the load-bearing detail the framing keeps at the edge, and it is worth putting in the centre because it changes what the proposal is actually a proposal for.

OpenAI and Anthropic can be “rejected by the market” in a way Google and Meta structurally cannot. They are, in the balance-sheet sense, single-product companies wagering the entire enterprise on the compute-and-scaling bet paying off before the money runs out. If markets reject them, that means something specific and mundane: the capital raises stop clearing, the compute bills come due, and there is no ad business or search monopoly sitting underneath to absorb the fall. Google’s AI arm could underperform for a decade and Alphabet would still print money from search. Meta could burn through the entire Llama roadmap and the ad engine keeps running. Nationalization-on-market-rejection is therefore a mechanism that can only ever be triggered by the two firms whose corporate structure exposes them to that risk in the first place - which means the safeguard is calibrated to balance-sheet fragility, not to capability, not to alignment risk, and not to the “steering society” stakes the proposal is framed around.

The plain question, then, is what the Schneier-Sanders remedy actually protects against. If the fear is that AI development driven by corporate profit motives could steer society toward catastrophic unsafe outcomes, that fear applies with at least equal force to Google and Meta - arguably more force, since their AI divisions answer to shareholders who have other profitable businesses to protect and therefore weaker incentive to slow down for safety’s sake, not less. A nationalization trigger keyed to market rejection reaches the two companies who could plausibly go under and leaves untouched the two who cannot be forced into the one condition that activates the remedy. The safety net catches the startups. The conglomerates stay exactly where they are, immunized by the very diversification that makes them harder to regulate through crisis.

None of this makes Schneier and Sanders wrong that the US government needs some lever here. It makes the lever they have reached for a bailout instrument wearing an alignment argument’s clothes - built for the day OpenAI or Anthropic cannot make payroll, not for the ongoing fact that four companies, only two of which are named, are already making the decisions the essay says should belong to the public.

There is real work behind naming OpenAI and Anthropic specifically - it is honest about which firms could actually fail, and failure is the only door through which government control walks in cleanly, without a fight from shareholders who are still winning. The trouble is what does not walk in behind it. A safety proposal that only bites when the balance sheet bites is a proposal that lets the firms with the strongest balance sheets keep steering, unbothered, while the argument’s real target - corporate incentive over public interest - quietly survives the rescue of its two least-protected competitors.