Oriole Networks announced continued progress in its collaboration with AMD, in support of the UK’s Advanced Research & Innovation Agency (ARIA) Scaling Inference Lab.
The work brings together Oriole’s photonic networking system and AMD Instinct GPUs and AMD EPYC CPUs to demonstrate how next-generation network fabrics address the growing performance, latency, and energy constraints of AI infrastructure.
In the collaboration, which has been underway for more than a year, Oriole is set to deploy the world’s first pure photonic AI network at scale, built to supercharge AI performance at the system level by providing the lowest possible latency.
Oriole is contributing its PRISM photonic networking solution, which replaces electronic switches in the network core with nanosecond-scale optical circuit switching. AMD is supporting the programme by providing CPU and GPU hardware, and technical collaboration to develop and run large-scale network models relevant to frontier-scale AI systems.
Madhu Rangarajan, Corporate Vice President, Compute and Enterprise AI Business, AMD, said: “AMD is excited to collaborate with Oriole on the ARIA Scaling Inference Lab cluster. Oriole’s AI backend networking with nanosecond optical circuit switching represents a fundamentally different way to connect accelerators at scale. We are helping to validate how photonic fabrics can work alongside AMD compute to deliver the low-latency, high-bandwidth connectivity that AI Inference workloads demand.”
It also marks the first commercial deployment of Oriole’s technology, which has gone from R&D to production in just three years. Oriole’s xPU-agnostic designs are now locked and set for wider rollout across the industry in 2027 to meet growing demand from multiple accelerator platforms.
James Regan, CEO of Oriole, said: “A year ago, we were proving the physics; today, we’re proving the business. Our collaboration with AMD has moved from concept to deployment to a system an order of magnitude larger, and the data proves this is already driving performance increases at pace. This is what it looks like when photonic networking stops being a research curiosity and starts being the foundation of how serious AI infrastructure gets built.”
Reimagining AI networks
At the core of the network is Oriole’s technology, PRISM: the world’s first AI networking platform that routes data as photons rather than electrical signals. For decades, data centre networks have run on electrical switches that are inefficient, power-hungry, and generate enormous heat. Coupled with the rise of AI, with its need for thousands of chips exchanging data trillions of times per second, data centre networks have been pushed to the breaking point.
PRISM removes the need for electronic switches entirely, replacing them with nano-second-switched optical circuits, which cuts core power consumption by 81%. With photons able to travel directly from chip to chip, GPU idle time drops from 60% today to less than 1%. With less hardware in the loop, PRISM can also reduce dependency on the complex supply chain that underpin today’s networking hardware and can minimise the need for cooling, thus slashing water usage. As the work with AMD shows, this leads to supercharged AI output with more tokens per second and more users served simultaneously from the same hardware.
“In many ways, what Oriole is doing in photonic networking mirrors what companies like Cerebras achieved on the compute side: a fundamental rethinking of how AI systems are architected. But this time, the focus is on the network. It is not just about a new technology entering the market, but a broader paradigm shift where incremental improvements are no longer sufficient.
“The next phase of AI infrastructure will be defined by architectures built natively around the physics of data movement, rather than working against it. Photonic networking enables that shift, unlocking the step-change in bandwidth, latency, and energy efficiency required to support the next generation of AI systems,” Regan told Electronic Specifier.
Crucially, PRISM is not built for any single chip vendor; it works across any accelerator platform, giving the wider industry a path to frontier-scale system-wide performance without the need for proprietary stacks.
Suraj Bramhavar, Program Director at ARIA, said: “Meeting the demands for modern AI requires rapidly identifying ways to improve the performance and cost-efficiency of large-scale AI clusters. ARIA is thrilled to collaborate with Oriole and AMD to demonstrate the benefits of this new technology and it’s exactly the type of collaboration, between innovative startups and industry leaders, that the Scaling Inference Lab was designed to foster.”
