Growing AI power slurpage prompts MPs to examine low-energy computing
Committee launches inquiry into emerging chip designs to curb datacenter energy use
MPs are probing whether radically different, low-energy chip designs can stop AI from turning the UK's power grid into a bottleneck.
The UK's Science, Innovation and Technology Committee has opened a short inquiry into whether so-called low-energy computing could rein in the ballooning electricity demands driven by AI, as model sizes and data volumes continue to scale in ways that make even hyperscale datacenters look modestly power-hungry.
Datacenters already account for about 2.5 percent of the UK's electricity use, with demand expected to quadruple by 2030. That puts pressure on a government trying to hit net-zero targets while simultaneously backing AI as a growth driver.
At the center of the inquiry is a cluster of emerging technologies that sound more like lab work than a planning application. Neuromorphic computing, which mimics the way the human brain processes information, and silicon photonics, which uses light instead of electrons to move data, are being pitched as ways to slash the energy cost of computation. Put the two together, and you get neuromorphic photonics – a nascent approach that researchers claim could deliver far more compute per watt than today's silicon.
The committee first got a look at these ideas during its "Under the Microscope" exercise, where researchers effectively auditioned their work for parliamentary scrutiny. Now MPs want to know whether any of it can move from promising prototype to something that actually takes pressure off the grid.
Dame Chi Onwurah, who chairs the committee, didn't sugarcoat the problem. With datacenter demand on track to surge, she said the UK faces a "fundamental question" about how to scale AI infrastructure without overwhelming the electricity system, particularly as the 2030 clean energy deadline looms.
"Through this inquiry, we'll examine whether emerging approaches like neuromorphic computing and silicon photonics can help meet our current and future energy and compute demands. We'll explore the UK research and innovation underpinning these technologies and assess our sovereign capabilities in low-energy computing," she said.
- Artificial brains could point the way to ultra-efficient supercomputers
- Bezos-backed Unconventional AI aims to make datacenter power problems go away
- India has satisfied its supercomputing needs, but not its ambitions
- Cyborg dreams move closer to reality with low-power artificial neuron
"It's vital for the UK's technological future that we expand our energy capacity, but it should not be at the cost of sustainability. I encourage anyone with expertise in this area to submit evidence to the committee."
The inquiry will dig into how mature these low-energy approaches really are, how long it would take to deploy them at scale, and whether the UK has anything resembling sovereign capability in this space or is likely to end up importing the tech along with the power bill.
It will also examine what the government is doing to support the research, and whether that backing reflects the scale of the problem. MPs want to know if these technologies can be deployed in practice – or if they remain safely confined to conference slides. ®
