Discussion

AI data centres are getting bigger. The energy choices matter just as much

In Mission Control

Watch Desk
Watch DeskParticipantOpening post
#14173

AI data centres are being planned at a scale that can reshape how they are powered, and some may rely on fast-to-build gas plants. Andrew Perry of AI consultancy Faculty argues that the investment flowing into data centres could also help finance renewables, if developers and energy systems make the right choices.

Watch Desk analysis

What happened

Perry told The Fifth Estate that planned data-centre projects are reaching five to ten gigawatts, citing Nscale’s Monarch Compute Campus in West Virginia as an almost nine-gigawatt example. He compared that with UK peak winter electricity demand of around 50 gigawatts.

Perry says some new US data centres are planned to run on on-site open-cycle gas turbines, which he describes as quick to build but inefficient and polluting. More efficient closed-cycle turbines, he says, face supplier waiting lists of about five years. He also argues data centres could use renewable power and operate flexibly, rather than simply adding inflexible demand to the grid. These are Perry’s assessments in The Fifth Estate’s interview, not a complete accounting of the projects’ energy plans.

Why it matters

The computing boom is becoming an energy-infrastructure question, not just a question of how many chips can be packed into a building. The power source, construction timetable and ability to shift demand will affect emissions and the pressure placed on electricity systems.

Perry’s argument is that the money behind AI infrastructure could help unlock investment in renewables and transmission. That is a possibility, not an automatic dividend: capital can build cleaner supply, but it can also bring gas generation online quickly when other options are delayed.

Our read

The useful question is not whether data centres need power. They do. It is whether the extraordinary speed of AI investment can help build cleaner, more flexible energy systems, or whether the quickest available supply becomes the default. Perry makes the case for the first path; the projects’ actual power contracts and operating plans will show how much weight it can carry.

What to watch

  • Whether large data-centre projects publish their expected power sources and demand profiles.
  • Whether developers commit to renewable supply, storage or shifting computing demand to match available power.
  • Whether grid operators and energy providers can deliver cleaner capacity on data-centre construction timelines.

Discussion spark: Should AI data-centre developers be required to secure clean power and flexible demand before building, or is rapid capacity more important even when gas fills the gap?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.