UK pledges £1.1bn to build domestic AI computing capacity
The UK government announced £1.1bn in AI hardware funding on June 8, with £750m allocated to a national AI supercomputer expected to be operational by 2030. The investment forms part of the broader £500m Sovereign AI programme launched by the Department for Science, Innovation and Technology in March.
Of the supercomputer funding, £400m will go directly to chip procurement. The government will split this between £150m for next-generation inference chips and £250m for specialised processors as AI technologies mature.
The chip spending includes a direct purchasing commitment from UK-based companies. Prime Minister Keir Starmer framed the approach with the phrase "start here, scale here and stay here" - an explicit attempt to keep AI firms operating in Britain rather than relocating.
The supercomputer will integrate advanced processors and specialised AI accelerators with existing facilities including Isambard-AI and Dawn.
Private sector follows with major commitments
AMD pledged up to £2bn and cloud provider Nebius committed £1.7bn to expand AI capacity in the UK, both announcements made during London Tech Week.
Scottish infrastructure firm Argyll Data Development launched what it describes as the UK's first sovereign AI inference cloud, built with chip designer SambaNova. The platform gives UK organisations access to high-performance inference without routing data through overseas infrastructure.
Energy constraints shape infrastructure choices
Energy consumption has become a critical constraint for sovereign compute projects across Europe. Regulators are introducing measures to curb data centre power use and reduce pressure on national grids.
SambaNova's Reconfigurable Dataflow Unit architecture operates within existing UK data centres without requiring liquid cooling or specialist power infrastructure. This design choice could materially reduce the energy overhead of moving AI inference onshore.
Peter Griffiths, chairman at Argyll Data Development, said the platform offers "premium AI inference running entirely within UK borders, at production scale, on existing data centres."
Rodrigo Liang, CEO at SambaNova, added that high-performance inference typically demands "liquid cooling, specialist power, and purpose-built facilities" - requirements that make sovereign deployment slow and expensive. Argyll's deployment sidesteps these constraints for the first time in the UK.
Learn more about AI for Government initiatives and infrastructure strategy.
Your membership also unlocks:
