‘We can’t stifle it’: Trump says AI leaders ‘have to self-police’ AI technology that ‘will be bigger than the industrial revolution’

President Trump said leading AI companies signed a voluntary accord involving internal and external technology reviews, while pledging that the U.S. government would not constrain AI development. The White House meeting included executives from Nvidia, OpenAI, Anthropic, Amazon, Google, Meta, Microsoft and X, signaling continued policy support for domestic AI investment. Companies also reportedly agreed to address local opposition to data centers through support for schools and lower energy costs, while Trump framed AI as a strategic U.S.-China competition.

Analysis

The investable implication is not a broad software rerating; it is a lower near-term probability of federally imposed capability caps that would interrupt hyperscaler capex and NVIDIA’s accelerator order cadence. NVDA remains the cleanest beneficiary, while MSFT, AMZN and GOOG gain through reduced regulatory uncertainty around monetizing enterprise agents and cloud AI services. The voluntary-review framework is economically preferable to binding pre-deployment approval, but absent published standards it should not yet be credited as a durable reduction in legal or compliance costs.

The second-order constraint shifts to physical infrastructure. Efforts to defuse local data-center opposition through community payments or lower power costs effectively raise the all-in cost of incremental compute; this favors scaled operators with contracted generation, transmission relationships and balance-sheet capacity over smaller GPU-cloud providers. Utilities and gas-fired power supply in key data-center regions could gain, while hyperscaler margins face a modest 6-18 month headwind if power concessions become a normalized permitting cost rather than a one-off local mitigation expense.

Near term, this is a sentiment support for AI capex rather than an earnings catalyst. The 1-3 month risk is that voluntary commitments prove insufficient after a visible model-safety failure, creating a bipartisan push for mandatory audits, liability rules or deployment restrictions; that outcome would compress the AI-premium multiples of META and NVDA most sharply. A less appreciated downside is China policy: a more explicitly zero-sum posture increases export-control tightening risk, which can impair NVDA’s addressable market even as domestic demand stays strong.

Key Decisions for Investors

• Maintain/accumulate NVDA on broad tech pullbacks rather than chase a policy headline: the thesis is preserved hyperscaler capex through the next two earnings cycles, with a 6-12 month horizon. Falsify on evidence of materially lower cloud-provider AI capex guidance or renewed China revenue restrictions; use a 10-12% drawdown from entry as risk review level.
• Pair long MSFT / short META over 1-3 months: MSFT has enterprise distribution and Azure monetization that are less exposed to consumer-facing safety or data-use backlash, while META’s AI valuation support is more dependent on sustained infrastructure spending without directly monetized cloud revenue. Exit if META demonstrates a clear AI-driven acceleration in ad pricing or MSFT Azure growth decelerates materially.
• Watch, do not yet buy, regulated power exposure such as CEG, VST and selected data-center utilities: local mitigation requirements could support power-contract pricing, but the missing data are regional load commitments, interconnection queues and who bears concession costs. Initiate only after disclosed long-duration data-center power contracts or confirmed permitting-linked rate support.
• Hedge AI-beta exposure with NVDA put spreads dated 3-6 months if implied volatility is favorable, focused around the next major policy or earnings window. The relevant tail event is a safety incident followed by binding federal rules, not the current voluntary language; limit premium to a small fraction of underlying AI gains.