The most durable part of the AI buildout may not be the chips at all, but the wires, substations and power plants behind them. A portfolio manager at the Ecofin Global Utilities & Infrastructure Trust argues that electricity demand, after a quarter-century of stagnation, could climb roughly 50% over the next decade, and that equity markets have not fully priced in what that means for regulated and contracted utilities.
The backdrop is a grid built largely in the decades after World War II and starved of investment since. Washington's 2022 Inflation Reduction Act and Germany's €500 billion infrastructure programme are early attempts to pay for the catch-up. Meanwhile, data-centre operators are locking in power supply through multi-decade procurement contracts, which hands generators and network owners something investors rarely get in this sector: long-dated, visible cash flows tied to a fast-growing customer.
The numbers the manager cites show why the setup looks interesting on valuation. Brookfield Renewable, down more than 20% from its late-May high, is described as expecting roughly 15% annual earnings growth over three years, with about 90% of revenue contracted and 70% inflation-linked, alongside a yield near 5.5%. Greece's grid operator ADMIE trades around 14 times earnings with a 2.7% yield, backed by a regulated return on equity of 9.8% and a plan to double its asset base by 2029. Scotland-based SSE, at under 12 times earnings, expects 80% of revenue to be regulated or contracted by 2030 and a £33 billion funded investment programme.
A few cautions belong in any honest reading. These are one manager's views, and the figures are forecasts, not guarantees. Regulated businesses live and die by the returns regulators allow, and heavy capital programmes are usually financed with debt and new equity, which makes the path of interest rates matter. Utilities are also not immune to a broader AI wobble if data-centre demand disappoints, although contracted revenue offers more cushion than most of the technology trade.
So what for long-term investors? The pattern worth noticing is that the AI theme can be approached through businesses where growth comes with inflation protection, regulated returns and a dividend that pays you to wait. When the market is nervous about AI spending, the companies selling the picks and shovels of power and transmission, rather than the models themselves, deserve a closer look on fundamentals: how much revenue is contracted, how generous or stingy the regulator is, how much debt funds the build, and whether the starting valuation leaves room for error. Patient capital tends to be rewarded in exactly this kind of slow, essential, unglamorous infrastructure.