Peter Thiel's macro fund has re-entered U.S. equities with Amazon as its largest disclosed position, but the rest of the portfolio reveals a more significant wager on the infrastructure that powers artificial intelligence.
Thiel Macro's latest 13F filing, covering the period ending June 30, shows $418.7 million in U.S.-listed holdings, a sharp reversal from the previous two quarters when the fund reported no equity positions. Amazon leads the list at $118 million, representing 28.2% of the portfolio.
However, the remaining holdings paint a clearer picture. The fund holds stakes in Vista Energy, Vistra, American Electric Power, DTE Energy, FirstEnergy, CMS Energy, and X-Energy. Combined, these energy and power-related positions account for roughly 72% of the disclosed value, according to the filing.
Amazon as a direct AI play
Amazon offers a straightforward route into the AI infrastructure boom. Its cloud division, AWS, reported a 37% year-over-year revenue increase to $42.2 billion in the second quarter, marking its fastest growth in 18 quarters. The company also raised its 2026 capital expenditure guidance to about $220 billion, underscoring its aggressive push to expand cloud and AI capacity.
Morningstar senior equity analyst Dan Romanoff noted after the results that "the surging demand spans both traditional and AI workloads," adding that the strength supports Amazon's massive investment program. Deutsche Bank analyst Benjamin Black expects Amazon's capital expenditure to approach $300 billion in 2027, as reported by MarketWatch.
Amazon has stated that AWS expansion requires spending ahead of demand on land, power, buildings, chips, servers, and networking equipment. This aligns with the broader market narrative that AI's growth is increasingly constrained by energy availability.
The energy-heavy portfolio
After Amazon, the fund's largest positions are Vista Energy at $75.9 million, Vistra at $59.1 million, and American Electric Power at $42.2 million. DTE Energy, FirstEnergy, CMS Energy, and X-Energy round out the list.
Not every holding is a direct AI bet. Vista Energy, for instance, is an oil and gas producer focused on Argentina's Vaca Muerta shale formation. Yet the concentration across utilities, power generators, and energy producers makes the infrastructure theme difficult to ignore.
T. Rowe Price's Dom Rizzo told MarketWatch that hyperscaler capital expenditure could reach $1.5 trillion to $1.6 trillion in 2027. He believes companies like Amazon are approaching an acceleration in spending because returns on deployed AI capital remain attractive.
Power is becoming a critical bottleneck. Melius Research analyst James West, as reported by The Wall Street Journal, sees an advantage for existing generators because expanding operating plants can be cheaper and quicker than building new facilities. He suggested that major data-center supply agreements could emerge "any day now."
Risks in the power trade
That interpretation requires caution. Thiel has not publicly described the portfolio as an AI-energy strategy, and a 13F filing reveals only holdings, not the investment thesis behind them.
The power trade is also becoming more complex. The Wall Street Journal reported that generators like Vistra have faced pressure as regulators in Texas and the PJM market respond to higher electricity costs and surging data-center demand. Measures encouraging new generation could eventually weaken the scarcity economics that benefit incumbent plants.
Amazon faces a similar risk. Its AI opportunity is enormous, but management must still demonstrate that record infrastructure spending will generate sufficient returns. As the market watches, Thiel's portfolio suggests a bet on the physical layer of the AI boom—one that may be as important as the software and chips.
This article is for informational purposes only and does not constitute financial advice.
