Peter Thiel's Founders Fund disclosed an 8-stock equity portfolio in its Q2 13F filing, with the majority weighted toward utilities. The fund is not chasing Nvidia or hyperscalers. It is positioning for the constraint no one priced six months ago: power infrastructure. The thesis is simple—AI compute scales faster than transmission capacity, and utilities with permitting momentum win before the next chip generation ships.
The filing shows Founders Fund holding stakes across regulated and independent power producers, including names exposed to data center interconnection queues and industrial load growth in Virginia, Texas, and the Pacific Northwest. The fund exited or trimmed semiconductor exposure during the quarter. The move reflects a view that chip supply is a solved problem and that the marginal constraint on AI deployment is now measured in gigawatts, not TOPS. Utilities with existing co-location agreements or joint ventures with hyperscalers appear overweighted relative to the fund's historical sector allocations.
This matters because it isolates a second-order infrastructure choke point that most growth allocators are ignoring. Data centers require 15 to 30 megawatts per facility at minimum, and new builds targeting AI workloads are pushing 100 to 200 megawatts. Interconnection queues at regional transmission organizations now stretch 3 to 5 years for new capacity, and permitting timelines for generation assets remain a regulatory quagmire. Utilities with brownfield expansion capacity, existing substations near fiber routes, or pre-approved co-location frameworks have pricing power that does not yet show up in consensus models. The Founders Fund position suggests Thiel sees this as a multi-year arbitrage between grid buildout and compute demand, where the former determines deployment velocity for the latter.
The filing also surfaces a broader reallocation thesis: hyperscaler capex is no longer fungible. Microsoft, Google, Amazon, and Meta have each committed $50 billion-plus annually to AI infrastructure, but they cannot deploy it without firm power commitments. That shifts bargaining power to utilities with available capacity and creates a new class of infrastructure bottleneck that venture and growth allocators have underweighted. Founders Fund is early, but the portfolio construction is legible—own the firms that control access to the grid, not the firms that sell into it.
Operators should track co-location announcements from Dominion Energy, Vistra, and Talen Energy over the next 90 to 120 days, as well as any revised interconnection timelines from PJM and ERCOT. Family offices should watch for secondary utility positions appearing in September 13Fs from other tech-focused funds, which would confirm the thesis is spreading. Permitting reform legislation at the federal level, particularly around transmission siting, would materially accelerate the timeline and compress the arbitrage window.
The 8-stock portfolio is a distillation, not diversification. Thiel is betting that the AI infrastructure story pivots from silicon to substations, and that the market has not repriced the utilities that sit between hyperscalers and their next 100 gigawatts of committed load.