Oriole Networks, a UK startup which uses light to train large language models (LLMs), has raised £10m in seed funding as it hopes to solve AI’s critical energy problem.

The round was co-led by University College London (UCL) Technology Fund, Clean Growth Fund, XTX Ventures and Dorilton Ventures, with support from Innovate UK Investor Partnership. 

UCL spinout Oriole Networks was created in 2023 by four founders, professor George Zervas, James Regan, Alessandro Ottino and Joshua Benjamin.

The startup has found a way of using light to connect thousands of AI chips into what the company calls a “super-brain.” 

According to the company, the super-brain can be used to train LLMs a hundred times faster than traditional methods with just a fraction of the energy. 

Oriole Networks said it has the power to revolutionise time-critical tasks, increase AI adoption, and accelerate machine learning algorithmic progress in a low-carbon world. 

The startup said this will bring enormous gains for all companies working with AI but will particularly benefit data centres. 

Data centres play a critical role in the evolution of AI and are facing systemic problems and unsustainable power consumption due to increased demands. 

“AI computational needs are increasing by 10 times every 18 months,” said George Zervas, Oriole Networks CTO. 

“Collective data movement across the servers in the data centre becomes a bottleneck which in turn limits the training and inference completion time,” he added. “This requires a fundamental shift in the co-design of next-generation networked systems.”

GlobalData forecasts that the overall AI market will be worth $909bn by 2030, having grown at a compound annual rate of 35% between 2022 and 2030.

In the GenAI space, revenues are expected to grow from $1.8bn in 2022, to $33bn in 2027, a compound annual growth rate of 80%.