Wood Mackenzie lifts data‑center demand

Data centers are turning into a power story as much as a computing story. That is the real point of Wood Mackenzie’s latest update. The firm said on April 28 that U.S. demand for electrical equipment tied to data centers could more than triple by 2030, because AI campuses now need huge amounts of steady power and need it fast. Basically, the constraint is no longer just chips or buildings — it is transformers, switchgear, interconnection slots, and whatever can get electrons to the site in time. (woodmac.com) ### What did Wood Mackenzie actually change? Wood Mackenzie’s new framing is bigger than a normal demand bump. It now sees the U.S. data-center electrical equipment market rising from $20 billion to $65 billion between 2025 and 2030, with data centers taking as much as 40% of the whole U.S. electrical-equipment market in an accelerated case, up from under 2% in 2020. That is a structural shift, not a cyclical one. (woodmac.com) ### How big is the load growth? The headline number is 110 GW of U.S. data-center capacity by 2030, up from about 24 GW in 2026. Wood Mackenzie says that would account for 68% of total U.S. load growth over the period. It also says data centers would add more than 400,000 GWh of electricity consumption — about eight times the electricity demand growth tied to EVs in the same window. (woodmac.com) ### Why does 259 GW matter? That 259 GW figure is not Wood Mackenzie’s base case for 2030. It is the long-range accelerated-demand scenario for 2050, and that distinction matters because a lot of quick writeups are blurring the dates. The firm’s moderated case gets data centers to about 95 G(woodmac.com)y, more capital spending, modular builds, and behind-the-meter power. (woodmac.com) ### So what is broken right now? Power access. Wood Mackenzie says lead times for critical equipment are already running 18 to 36 months. It also says around 600 GW of projects in its U.S. pipeline are still looking for power capacity, versus 183 GW that have signed construction or electrici(woodmac.com)osed pipeline hit 241 GW. In plain English — demand is exploding faster than the grid can say yes. (woodmac.com) ### Is Wall Street seeing the same thing? Yes, though with a slightly different lens. Goldman Sachs said in December that global data-center power demand could rise 175% by 2030 from 2023 levels, driven by AI expansion and new data-center buildouts. The exact percentages differ across firms and updates, but the direction is the same — data centers are becoming one of the biggest new electricity loads on the system. (goldmansachs.com) ### Why are fuel cells suddenly in the conversation? Because waiting five years for the grid is not a workable plan if you are racing to deploy AI capacity. Wood Mackenzie explicitly points to behind-the-meter generation as one way developers could bypass grid and labor delays. Bloom Energy(goldmansachs.com)microgrid campus, and Bloom’s Q1 results two days later showed revenue up 130.4% year over year to $751.1 million. FuelCell Energy is making the same pitch from a smaller base, saying its data-center pipeline has grown sharply as developers prioritize faster time to power. (woodmac.com) ### What should you watch next? Watch the boring gear. Transformers, switchgear, ATS units, and PDUs are now the choke points. In hyperscale alone, Wood Mackenzie sees padmount transformers rising from 1,573 units in 2025 to 9,395 by 2030. That is the kind of number that tells you this boom is moving out of the software layer and into heavy electrical infrastructure. (woodmac.com) ### Bottom line The AI buildout is starting to look like an electricity buildout wearing a data-center label. Wood Mackenzie’s update matters because it says the winners will not just be chipmakers or cloud providers — they will also be whoever can deliver power, equipment, and time-to-grid fast enough. (woodmac.com)