From the Probability Desk: The Power Wall — can the grid build fast enough for the age of AI? For two years the constraint on AI was chips; increasingly it is electrons. This five-minute cinematic walks through the data-centre power crunch and what it means for the world's largest, longest-horizon investors.
Voiceover transcript
From the Probability Desk at Universal Asset Owners — this is The Power Wall. The story of the constraint that now sits between the world's largest investment in artificial intelligence and the electrons it needs to run.
The question this desk put a number on: will US data-centre electricity consumption reach three hundred terawatt-hours in calendar year 2027? From fifty thousand simulated paths, our answer is approximately sixty-eight percent yes.
For two years the constraint on artificial intelligence has been narrated as chips. It is increasingly electrons. Across 2025, the electricity drawn by AI-optimised data centres grew by roughly fifty percent. The binding bottleneck on the next leg of the build-out is no longer fabrication capacity or capital — it is the ability to connect new load to a power grid that takes more than five years to interconnect.
PJM Interconnection — the largest US grid operator — cleared its capacity auction at a record three hundred and twenty-nine dollars per megawatt-day, the ceiling of the approved price collar. The interconnection queue now stretches beyond five years. The International Energy Agency estimates roughly twenty percent of planned data-centre projects are at risk of delay because of grid constraints.
Every major hyperscaler has now signed at least one nuclear deal. Thirteen announced projects commit more than nine-point-eight gigawatts of nuclear capacity. Microsoft is restarting Three Mile Island. Meta has contracted up to six-point-six gigawatts across nuclear developers. Because reactors take years, roughly one-fifth of planned US data centres have begun site work for on-site natural-gas generation — building their own power plants rather than waiting for the grid.
This matters to universal owners more than almost anyone else, because the same institutions that hold the hyperscaler equities at index weight are also the dominant capital behind the physical build-out. Sovereign and pension funds committed an estimated one hundred and twenty billion dollars to AI infrastructure across 2025 and 2026. A long-horizon owner is therefore exposed to this story twice — once through the companies that need the power, and once through the wires, turbines, reactors, and land that must deliver it.
The Probability Desk weights four scenarios. Build-out delivers — thirty-five percent — the grid, gas and nuclear self-supply broadly keeps pace, power is scarce but served. The Power Wall binds — thirty-five percent — interconnection and equipment shortages cap realised load below demand; power prices spike; some compute is stranded. AI demand disappoints — fifteen percent — the flat-era replay. Acceleration overshoot — fifteen percent — demand and supply both surprise high.
What the market is still under-pricing is duration and delivery. The assets that earn from scarce, dispatchable, grid-connected power over a decade — independent power producers, nuclear and gas-fired baseload, grid equipment, and the long-dated infrastructure debt that funds it — are priced for a cyclical surge, not a structural decade-long capacity shortage.
The mirror-image risk: the demand forecasts are wrong again, as they spectacularly were after 2007, when US electricity demand was essentially flat for fifteen years despite strong economic growth. The flat-era cautionary record insists we keep that tail on the board.
For a universal owner, the investable question is not whether to own the hyperscalers — you already do, at index weight. It is whether you own the electrons that the hyperscalers have discovered they cannot do without. The Power Wall, from the Probability Desk at Universal Asset Owners.