Oriole, a UCL spinout attempting to train large language models (LLMs) faster and less energy-consuming than the current standard, has raised £10m.
The startup was founded in 2023 in response to the growing concern in the AI sector over the extreme energy requirements of running generative AI models.
Oriole claims to have found a solution by using light to connect thousands of AI chips together. The startup said the connected GPUs form what it calls a “super-brain”, which can then train advanced LLMs.
The spinout said training AI models with its super-brain, compared with the standard approach, is 100 times faster and uses a fraction of the energy.
“As the demand for compute continues to increase, it is critical to find new solutions that can address these challenges in a sustainable and carbon efficient manner,” said James Regan, CEO of Oriole.
CTO George Zervas added: “Collective data movement across the servers in the data centre becomes a bottleneck which in turn limits the training and inference completion time. This requires a fundamental shift in the co-design of next-generation networked systems.”
The seed funding round was co-led by UCL Technology Fund, Clean Growth Fund, XTX Ventures and Dorilton Ventures with additional support from Innovate UK Investor Partnership.
UCL Technology Fund last week backed a fellow spinout developing a new form of LLM AI model. Stanhope AI, which raised £2.3m, is attempting to create a human brain-like AI model using neuroscientific principles.
The post UCL spinout Oriole raises £10m for low-energy AI LLM training appeared first on UKTN.
