The AI Safety Menu

All laws

Public research compute

Public AI research compute. Government computing power for university and public AI research.

Lead costAlmost noneof America's lead over China, over 3 years
No p(doom) effectNonefrom a 5% starting estimate
Holds up○ Training○ Lab's own use○ Public release
Enact it on the menuSee the findings

What it does

Government provides computing power for university and public-interest AI research, through the National AI Research Resource and California's CalCompute.

Applies to: Researchers outside frontier labs.

Where things stand

The NSF pilot supported over 800 projects; an operations center was set up in September 2026. The CREATE AI Act passed a House committee 29–0 in June 2026. CalCompute operates only if the state funds it.

Why it costs almost no lead

It broadens access to AI research, but its scale is tiny next to frontier labs. Under a day.

Biggest unknown: Whether public research feeds frontier progress through ideas rather than compute.

Its effect on p(doom)

Gives universities computing power, some of it for safety research.

It could help academic safety research slightly.

The strongest case that it costs more

Academic research produced many of the ideas frontier labs now use. If public compute seeds the next big algorithmic advance, its value could be far larger than its chip count suggests.

The debate

For

Sources

Rough starting points, not precise forecasts. Lead costs assume China doesn't depend on U.S. models, the case least favorable to safety laws, and count 3 years. On the menu you can change every assumption and put in your own numbers. Last priced 2026-09-26.