Research & Commercial Insight — AI and the Long-Term Portfolio. UAO Research, June 7, 2026.
The question. The AI-infrastructure buildout is being marketed to long-horizon owners as the perfect long-duration asset: contracted, inflation-linked cash flows from creditworthy hyperscaler tenants, secured against physical plant, over a multi-decade demand curve — a textbook match for a 50-year liability. For most allocators that pitch is simply an opportunity. But for a universal owner — a fund so large and diversified it effectively owns a slice of the whole economy — is buying the data-centre build a diversifying allocation, or a concentrated doubling-down on an exposure the fund already carries?
What the evidence says. Start with the scale, because it changes the category. The four largest hyperscalers have guided to roughly $725 billion of capital spending in 2026, up about 77% from $410 billion in 2025 Tom's Hardware, 2026. That capital lands as physical electricity demand: the International Energy Agency projects global data-centre consumption roughly doubling from about 415 TWh in 2024 to around 945 TWh by 2030, with the AI-driven slice growing about 30% a year IEA, Energy and AI, 2025. The binding constraint of the cycle has migrated — from chips, to capital, to electrons. And the capital is increasingly institutional: on June 5, Blackstone- and CPPIB-owned AirTrunk committed more than $30 billion to build five gigawatts in India by 2030 AirTrunk, June 5, 2026. A public pension fund is now an equity owner of the compute backbone, not a distant lender to it.
Now walk that single exposure around a universal owner's portfolio, and watch it appear at four desks at once.
Desk one — you fund it. Through the infrastructure-equity and private-credit sleeves, the fund is an LP in exactly the AirTrunk-style vehicles being raised under the "golden age of infrastructure" banner. This is the allocation the fund chooses.
Desk two — you power it. Through utilities, grids, gas and the commodities that feed them, the fund owns the supply side of the ~945 TWh. The same demand that lifts the data-centre cash flows lifts the power-price exposure it already holds.
Desk three — you are repriced by it. Through index equities, the fund owns the hyperscaler tenants whose $725 billion of capex is the demand. Its passive book now rides a single capital cycle, whether or not it ever bought a data centre.
Desk four — you pay for it. Through its beneficiaries, the fund is exposed to the ratepayer. On the PJM grid, Q1 2026 wholesale power rose 76% year on year to $136.53/MWh — the largest jump in the grid's history — with the independent market monitor attributing 63% of the increase to data-centre load and roughly $9.3 billion in costs landing on ratepayers Utility Dive, 2026. The pensioners and citizens the fund serves are the voters absorbing that bill.
The disagreement. The genuine bull case is that patient, permanent capital is the natural financier of a real, generational infrastructure cycle, and that contracted hyperscaler leases are sound liability matches. That case is not wrong. The counter is twofold. First, return: Bain & Company estimates the buildout needs about $2 trillion in new annual revenue by 2030 and forecasts an $800 billion shortfall even on generous assumptions Bain & Company, September 2025 — a long-duration asset only matches a liability if the cash flows actually arrive, and some capacity is being built ahead of the revenue. Second, concentration: for a universal owner the four desks are not four bets but one, gated by a single constraint — power — that the financier does not govern. (Treat Bain's gap as a contested forecast, not a fact; pair it with each manager's own disclosures.)
What it means from the allocator's seat. The practical implication is not "avoid the trade." It is that the universal owner cannot treat AI-infrastructure as a discrete, diversifying line item, because it already owns the exposure four ways. The lever that remains is the one only a permanent, whole-economy owner holds: voice. Because the buildout's externality — higher power bills, grid strain, the political backlash — sits inside the portfolio, stewardship stops being an ESG overlay and becomes risk management for the whole book. That means pushing portfolio companies on capex discipline, pressing for grid investment and "bring-your-own-generation" rules at the tenants, and engaging on ratepayer fairness in the jurisdictions where the fund's beneficiaries live. The owner's edge was never the clever allocation; it is that it will still be here — owning the grid, the compute, and the consequences — in 2050.
What to watch next. Whether hyperscaler 2026 capex guidance holds through the year or is trimmed; the next PJM and regional capacity-auction prints and any state moves to re-allocate data-centre grid costs; interconnection-queue and grid-buildout timelines as the true rate-limiter; and how many more sovereign and pension funds follow CPPIB from lender to direct owner of the compute backbone.
Sources
- International Energy Agency, Energy and AI — Energy demand from AI (2025). https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai
- AirTrunk, "Blackstone-backed AirTrunk plans to invest US$30bn and +5GW in India," June 5, 2026. https://airtrunk.com/blackstone-backed-airtrunk-invest-us30bn-5gw-india/
- Tom's Hardware, "Big tech's AI spending plans reach $725 billion" (2026). https://www.tomshardware.com/tech-industry/big-tech/big-techs-ai-spending-plans-reach-725-billion
- Utility Dive / Monitoring Analytics, "Data centers were a record share of PJM capacity costs" (2026). https://www.utilitydive.com/news/data-centers-pjm-capacity-auction/808951/
- Bain & Company, 6th annual Global Technology Report (September 2025). https://www.bain.com/about/media-center/press-releases/20252/$2-trillion-in-new-revenue-needed-to-fund-ais-scaling-trend---bain--companys-6th-annual-global-technology-report/
Continue the briefing. Read the daily brief · watch the daily video briefing · listen to The Universal Owner · view the chart of the day.
Produced and edited by the UAO editorial desk. Not investment advice.