Scaling Compute AI

Fund details

A key advantage of the digital electronics industry, constantly increasing computing power at lower costs, is disappearing. This coincides with a surge in demand for processing power due to AI. The traditional methods for training AI are expensive due to limited approaches and hardware which has serious economic, geopolitical and social consequences.

ARIA see an opportunity to draw inspiration from natural processing systems, which innately process complex information more efficiently (on several orders of magnitude) than today’s largest AI systems.

The goal is to increase and open up new vectors of progress in the field of computing by reducing the cost of AI hardware.

The programme seeks R&D creators, individuals and teams that ARIA will fund and support.

They will focus on three key tasks:

• Develop software simulation tools capable of estimating the performance of novel compute hardware focused specifically on training large-scale AI models
• Demonstrate working hardware meeting specified building block targets
• Demonstrate through systems-level analysis that if these building blocks were developed at scale, they could meet disruptive programme targets

Eligibility

This round consists of three technical areas including:

• TA1 Bold Solutions – project costs ≤£26m
• TA2 Bold Ideas – project costs ≤£8m
• TA3 System-level Software Simulation – project costs ≤£5m