The White House circulated draft AI framework proposals to OpenAI, Anthropic, and Google this week. The deadline is August 1.
Sam Altman spent July 29 and 30 on Capitol Hill. Demis Hassabis published his own counter-proposal. Neither man agrees on what a regulated AI market should look like.
The IAEA vs. the FINRA
Altman's model: a U.S.-led international forum modeled on the International Atomic Energy Agency. Countries and companies get certified. Access to frontier models and U.S. markets serves as the compliance lever. Those who refuse the framework lose access.
Hassabis's model: a FINRA equivalent for AI. Industry-funded, federally overseen, starting with voluntary pre-release reviews. The voluntariness is the key word: the proposal explicitly contemplates hardening into mandatory market-access rules over time.
Both men represent companies that benefit structurally from either outcome. The regulatory moat builds itself.
The Market Signal
The same week both CEOs arrived in Washington, Situational Awareness LP, the roughly $20 billion hedge fund built by former OpenAI researcher Leopold Aschenbrenner on an AGI infrastructure thesis, sold its entire public equities book in a single block trade before market open on July 30. The buyer was Citadel.
Long positions in CoreWeave, Bloom Energy, and listed crypto miners absorbed the July AI selloff until they could not. The fund still holds its Anthropic shares. In a July 24 letter to investors, Aschenbrenner called the selloff one of the best buying opportunities since early last year.
The conviction remains. The portfolio did not.
The Constraint That Neither Proposal Addresses
Bearish risks are structural. A mandatory pre-release vetting regime concentrates frontier AI development among the three companies already at the White House table. Open-source alternatives face existential regulatory risk under either framework. Smaller startups without legal infrastructure lose access to the largest markets.
Gartner estimates global data center electricity consumption will reach 565 TWh in 2026, up 26% year-over-year. Aging U.S. power grids are already under strain from AI demand. Energy infrastructure is the physical rate limiter for any AI capex thesis, regulated or otherwise.
The Fed held at 3.5% to 3.75% on July 30, a 9-to-3 vote with three dissenters pushing for a hike. Bitcoin stayed flat at $64,000 as equity markets fell 1.5%. The policy paralysis Lyn Alden described on MacroVoices is now the operating environment.
August 1 is both a White House deadline and the day Aschenbrenner opens for fresh capital. One of them will be proven right.
Follow @UTXOMacro for daily Bitcoin, macro, and AI breakdowns.