MPs probe low-energy chips as AI's appetite threatens the grid

What's happening

MPs on the Science, Innovation and Technology Committee have launched a short inquiry into whether radically different, low-energy chip designs can stop AI from turning the UK's power system into a bottleneck. It has been reported that datacentres already account for about 2.5% of the UK's electricity use and that demand could quadruple by 2030 — a projection that turns polite concern into something closer to panic for policymakers juggling net-zero targets and an AI growth agenda.

The tech on the table

At the centre of the inquiry are neuromorphic computing and silicon photonics — and the slightly sci‑fi sibling, neuromorphic photonics. It has been reported that researchers claim these approaches could deliver far more compute per watt than today's silicon; allegedly, combine brain‑like architectures with light to move data and you might slash energy bills. Promising? Sure. Ready to shoulder the nation’s compute load? That is precisely what MPs want to test.

Why it matters

Dame Chi Onwurah, who chairs the committee, framed it bluntly: how do we scale AI infrastructure without overwhelming the electricity system as 2030 looms? The inquiry will probe maturity, deployment timelines, UK sovereign capability, and whether government support matches the scale of the problem. Think of it as a reality check after the “Under the Microscope” auditions where researchers pitched prototypes to Parliament.

Next steps

MPs have invited evidence; the clock is short. Will these lab curiosities become industrial workhorses, or remain clever slides for conferences? Either way, the outcome will shape whether Britain builds more power capacity, imports tech (and kilowatt‑hours) — or bets on a new generation of chips to keep the lights on while the AI party rages.

Sources: The Register