Gang of 3: Zuckerberg, Musk, and Huang Call Trump to Oppose AI Regulation

Meta's Mark Zuckerberg, SpaceX's Elon Musk and NVIDIA's Jensen Huang reportedly persuaded President Trump to retreat from a proposed federal AI oversight body, reducing the near-term risk of mandatory pre-deployment frontier-model testing. The companies are escalating AI infrastructure investment: Meta guided 2026 capex to $130B-$145B despite quarterly free cash flow falling to $784M from $8.55B, while SpaceX directed $15.83B of $18.37B quarterly capex to AI compute. NVIDIA reported Q2 FY2027 revenue of $96.22B, up 105.85% YoY, guided Q3 revenue to $108.0B, and cited demand well above available supply, supporting continued spending across AI power, cooling and networking supply chains.

Analysis

The investable read-through is less about federal rulemaking than about preserving deployment velocity: hyperscaler and sovereign compute budgets remain the marginal demand driver for NVDA, ANET, VRT, ETN and GEV. A lighter federal posture may also reduce the probability of a costly pre-deployment approval delay, but it does not eliminate state-level liability, copyright, privacy, export-control or power-permitting constraints. The more important 6-18 month bottleneck is grid interconnection and data-center energization; that shifts incremental economics from GPU vendors toward electrical equipment and thermal-management suppliers once accelerator availability normalizes.

NVDA's demand visibility supports estimates near term, but large supply obligations are not equivalent to booked, non-cancellable revenue; customer financing, power availability and rack-level integration can defer recognition. The market is likely underpricing a 1-3 month rotation risk if lead times improve: buyers that have over-ordered scarce accelerators could redirect capex toward networking, storage, power and inference optimization, narrowing NVDA's scarcity premium without breaking the AI spend cycle. Conversely, sustained supply tightness plus continued hyperscaler capex guidance would keep estimate revisions positive.

META is the more nuanced loser/winner. Faster model deployment can improve ad targeting and engagement, but its valuation will increasingly hinge on whether monetization scales before depreciation, energy and financing costs absorb cash generation; premium-priced excess compute capacity is not a durable earnings stream without contracted demand. A voluntary federal framework could be constructive for META versus smaller model developers, but a fragmented state regime would favor the largest balance sheets and legal teams while increasing operating costs across the ecosystem.

Key Decisions for Investors

• Maintain/enter long NVDA versus short SOXX for a 1-3 month earnings-revision trade; use a 8-10% NVDA relative-underperformance stop. Thesis is falsified by a material reduction in next-quarter data-center guidance, disclosed order cancellations, or evidence that supply lead times are normalizing faster than demand.
• Add a 6-12 month basket long VRT, ETN and GEV, sized smaller than semiconductor exposure, to capture the power-and-cooling bottleneck. Favor staged entries after earnings because data-center revenue timing depends on project energization; exit if bookings/backlog conversion weakens for two consecutive quarters.
• Use META as a relative short against GOOG over the next 1-3 months only if management does not provide measurable AI monetization or raises capex again while free-cash-flow conversion deteriorates. The pair is invalidated if META demonstrates ad pricing/engagement acceleration sufficient to offset incremental depreciation and infrastructure expense.
• Do not position solely on the reported policy intervention. Set an alert for a federal executive action, agency rulemaking timetable, or state AI-liability legislation: a binding, model-testing regime could paradoxically favor GOOG and META through compliance scale while reducing the addressable market for smaller AI platforms.