Oriole Networks unveils photonic technology to revolutionise AI efficiency

In a bid to revolutionise artificial intelligence (AI) infrastructure, UK-based startup Oriole Networks has announced a groundbreaking technology aimed at reducing energy consumption and improving the efficiency of AI systems. The London-based company is leveraging advancements in photonics to develop networks of AI chips capable of training large language models (LLMs) with remarkable speed and energy efficiency.

Founded on two decades of research conducted at University College London (UCL), Oriole Networks is pioneering an approach that utilises light to address current shortcomings in AI processing. The company’s innovative use of photonic networking technology allows AI systems to train LLMs up to 100 times faster than conventional methods, while simultaneously cutting down on power usage significantly.

The leadership of Oriole Networks, under the CEO James Regan, is spearheading this effort. Regan articulates the company’s ambition to transform the AI sector by creating an ecosystem that tackles existing processing bottlenecks and promotes competition in the graphics processing unit (GPU) market layer. “Building on decades of research, we’re paving the way for faster, more efficient, more sustainable AI,” stated Regan.

The technological foundation of Oriole Networks is attributed to the pioneering work of UCL researchers, including Professor George Zervas alongside scientists Alessandro Ottino and Joshua Benjamin. Their patented intellectual property forms the core of the startup’s cutting-edge solutions. This innovation has garnered the interest of investors eager to support the development of sustainable AI technologies which meet today’s high-performance computing demands.

Ian Hogarth, a board member of Oriole Networks and partner at Plural, the firm that led the latest funding round, expressed confidence in the company’s potential. According to Hogarth, the application of advanced photonics research by Oriole Networks confirms the presence of vast untapped innovation within the AI sector. He highlighted the team’s expertise in both business development and the commercialisation of deep science, which is facilitating a fundamental shift in the design of next-generation networking systems.

Oriole Networks is currently planning the release of its first batch of products by 2025, marking a significant milestone in its journey towards reshaping AI’s infrastructure. This release is expected to significantly enhance data centre operations by reducing latency and lessening the environmental impact associated with energy-intensive AI tasks.

As the field of artificial intelligence continues to rapidly expand, the advancements being made by Oriole Networks showcase a promising direction for sustainable and efficient computing solutions.

Source: Noah Wire Services