Trump calls for plans to form federal ’AI Force’

President Trump said he plans to create a federal "AI Force" and appoint an AI czar, pledging not to impose statutory limits that could constrain U.S. artificial-intelligence development. He projected AI could eventually represent as much as 25% of U.S. GDP and framed maintaining leadership over China in AI, semiconductor supply chains and data-center infrastructure as a strategic priority. The pro-growth stance is supportive for major AI platforms and infrastructure providers, although it conflicts with calls from Anthropic, OpenAI, Google and Meta for a slower rollout of frontier models following autonomous-agent security incidents.

Analysis

The investable transmission mechanism is not a broad multiple expansion for frontier-model owners; it is reduced permitting and power-procurement friction for the physical AI stack. If federal policy pre-empts or pressures state/local opposition, the highest operating leverage sits with data-center electrical equipment and power suppliers—VRT, ETN, GEV and CEG—because incremental campus approvals convert into booked backlog and pricing power. SMCI benefits only if accelerated deployments remain server-architecture intensive; its exposure is more vulnerable to hyperscaler custom-rack substitution and component-margin competition than VRT or ETN.

Near term, the announcement is not independently monetizable until the appointee, legal authority, budget and permitting actions are known. A 1-3 month catalyst path is executive action affecting interconnection, federal land, transmission, environmental review, export controls or defense procurement; without these, this is chiefly sentiment support for GOOG and META rather than a change to revenue estimates. The more important 6-18 month effect is that faster construction raises already-tight transformer, switchgear, turbine and firm-power scarcity rents, while potentially worsening regional power-price and grid-reliability backlash.

Consensus may be over-crediting large platforms: lighter ex-ante safety rules reduce compliance cost, but also increase the probability of a high-profile misuse event that produces abrupt, model-specific restrictions and reputational costs. GOOG and META have balance sheets to absorb compliance and power capex, whereas smaller AI infrastructure firms face greater order-cancellation and working-capital risk if the policy push collides with power availability. APP is a weak read-through: advertising-model optimization has little direct sensitivity to data-center permitting, and should not be bought solely on this policy narrative.

Key Decisions for Investors

• Prefer a 3-6 month long VRT / short SMCI pair: VRT has cleaner exposure to electrical bottlenecks and backlog pricing, while SMCI embeds greater execution, gross-margin and customer-concentration risk. Reassess if hyperscaler capex guidance decelerates or VRT book-to-bill falls below 1.0x.
• Accumulate ETN and CEG on policy-driven pullbacks rather than chase platform beta; target a 6-18 month horizon tied to incremental data-center load and grid investment. Thesis is falsified by sustained hyperscaler capex cuts, material project cancellations, or state-level power-rate interventions that cap returns.
• Do not add directional GOOG or META solely on the announcement. Upgrade only after concrete federal actions reduce permitting timelines or after earnings show AI infrastructure spend translating into incremental monetization rather than higher depreciation; otherwise the principal near-term impact is additional capex burden.
• Set an event watchlist for the AI czar appointment, agency authority and any permitting/export-control order over the next 30-90 days. Absence of implementable action should fade the infrastructure-policy premium; a major autonomous-agent security incident would favor reducing high-multiple AI exposure and could widen VRT/SMCI relative performance.