Back to News
Scaling Inference Lab deploys first cluster to cut AI infrastructure costs
- The Scaling Inference Lab, delivered by CommonAI, has deployed its first cluster in partnership with Quettaflop AI. Cluster 1 will test whether older, depreciated GPUs can use intelligent software architecture to substitute for raw hardware capability, with the aim of delivering significantly cheaper LLM inference infrastructure and capability.
- Alongside this milestone, the Department for Science, Innovation and Technology has pledged an additional £20m+ in funding for the Lab, building on ARIA's initial £50m investment.
- Cluster 2 is planned for deployment in August 2026, in partnership with Callosum. Its mission is to explore heterogeneous hardware orchestration across diverse chip types, exploiting the individual benefits of each while working in harmony with one another, delivering more optimised hardware and efficiency for the wider UK AI ecosystem.
“There is an insatiable need to improve the cost efficiency of future AI infrastructure. While there is a lot of capital being deployed and a wide swath of technologies being explored to help meet this need, many are developed at the component level.
- Suraj Bramhavar, Programme Director, ARIA
Outside of the largest hyperscalers, there is comparatively little activity focused on the systems-level benefits of each underlying technology. ARIA has launched the Scaling Inference Lab, delivered by CommonAI to address this.”
To read the full announcement, see here.