Trump's 'morally binding' AI accord gives the six frontier labs the pen — and unblocks a $200B data-center pipeline

The document, formally the White House Accord on Super Intelligence: Joint Commitment on Frontier Responsibilities, was posted by Trump to Truth Social after the White House luncheon. It commits signatories to 'four layers of controls and audits' — internal safeguards on training and deployment, an internal oversight team, an independent external auditor, and an independent board committee — covering cybersecurity, biosecurity, and chemical-threat risk. It also states companies will 'meet regularly to establish standards and best practices' and notes that 'over time, it may make sense to codify these steps into laws or regulations.'

The signatories are six, not seven: Alphabet CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta Platforms CEO Mark Zuckerberg, OpenAI president Greg Brockman, xAI/SpaceX CEO Elon Musk, and NVIDIA CEO Jensen Huang. Microsoft and Amazon were present but did not sign. The accord is one page long; the Forbes copy of the text runs to 308 words.

Alongside the accord, Trump signed an executive order titled 'Inaugurating the Era of Super Intelligence.' Section 1 directs the executive branch to use 'Super Intelligence' and 'SI' in place of 'Artificial Intelligence' and 'AI' in official correspondence, public communications, websites, reports, and policy documents. Section 2 makes that direction binding within60 days for non-statutory documents. Section 3, critically, ties 'Super Intelligence' to the existing statutory definition of AI under 15 U.S.C. § 9401(3) and asks the Office of Science and Technology Policy, within 60 days, to propose a federal statutory definition that may 'modify, expand, or supersede' the current one.

The most consequential half of Trump's podium remarks was not about AI at all. With the Nov. 3 midterms five weeks out and 70% of Americans in Gallup polling opposing new AI data centers, Trump used the press conference to publicly endorse rapid data-center expansion and warned executives they must 'make massive contributions to communities' to neutralize bipartisan backlash. He told reporters that if local opposition continues, data centers 'will be forced to go overseas or other locations.'

Microsoft and Amazon attended the lunch but are not signatories. Their absence is the most telling feature of the day. Microsoft just completed the unwinding of its exclusive OpenAI cloud arrangement in April 2026 and is now routing internal workloads to its own models rather than paying Anthropic; OpenAI in turn spent2026 inking new Azure commitments through 2030 worth a reported $250B, even as Microsoft's Azure topped $100B for the first time in fiscal Q4 2026 (revenue $90.0B, +18% year-over-year). Amazon's AWS, the largest cloud by revenue, faces its own Jassy-vs.-Trump friction that was not resolved at the table.

The structural consequence: the six signatories — five of which are model developers and one of which is the dominant chip supplier — now own the standard-setting process. Anthropic's $965B post-money valuation and $65B revenue run rate (Series H, May 2026) and OpenAI's $852B valuation with a $40B run rate (March 2026 funding round) are now backed by a Trump-era 'in our image' framework rather than binding oversight. Smaller frontier-model startups and open-weight advocates — the constituencies that pushed back on a similar2023 Biden accord — were not invited.

The Sept. 29 deal layers an explicit White House endorsement over a buildout that hyperscalers and AI labs were already financing. NVIDIA's Q2 FY2027 (quarter ended Jul 26, 2026) printed $96.2B of revenue with $89.0B from Data Center, up 117% year-over-year — the segment now accounts for ~92% of company revenue. Meta Platforms raised2026 capex to $130B–$145B in August; Microsoft's capex-to-revenue was 34.9% TTM through June 30, 2026; and Oracle's capex-to-revenue hit 105% TTM through Aug. 31, 2026. CoreWeave, the only pure-play AI infrastructure stock that reports results, posted Q2 2026 revenue of $2.575B (+112% YoY) against a $104.2B contracted backlog and guided2026 capex to $35B–$39B.

On the power side, three listed names absorb the marginal kilowatt. Constellation Energy operates 32.4 GW of nuclear, gas, wind, solar, and hydro capacity and signed a 20-year power purchase agreement with Microsoft for the Three Mile Island restart in 2024. Vistra has 38.7 GW including nuclear, gas, coal, solar, and battery storage and is the largest U.S. independent power producer by capacity. NRG Energy operates 18 GW plus retail energy across six million customers. Equinix, the digital-infra REIT, runs $9.9B TTM revenue at a 65x trailing P/E — the market is already pricing AI tenancy into the colocation multiple.

The market was already pricing the AI infrastructure complex on volume, not on regulation. The Sept. 29 deal removes the binding-regulation tail risk that has lingered since 2023 and replaces it with an industry-friendly baseline. The structural winners are the AI utility complex (Constellation Energy, Vistra, NRG), the data-center landlords (Equinix), and the credit-leveraged pure-play builder (CoreWeave). The structural winners on the silicon side are NVIDIA — now both an accord signatory and the gating supplier — and the foundry ecosystem (TSMC, Applied Materials). Hyperscalers (Microsoft, Alphabet, Meta Platforms) get the cleanest regulatory clarity, with Oracle the most exposed to balance-sheet risk if capex monetization slips.