Oriole Networks says it will deploy a large-scale AI system using a pure photonic network with AMD Instinct GPUs and AMD EPYC CPUs for the UK Advanced Research and Invention Agency’s Scaling Inference Lab.
Oriole Networks said it will deploy a large-scale AI system powered by a pure photonic network for the UK Advanced Research and Invention Agency’s Scaling Inference Lab, using AMD Instinct GPUs and AMD EPYC CPUs.
The announcement, published by Oriole Networks and distributed through EIN Presswire, describes the project as an effort to address latency, performance, and energy constraints in AI data centers. Oriole says the system will use its PRISM technology alongside AMD hardware for ARIA’s Scaling Inference Lab.
AMD separately confirmed the collaboration in a press release about its commitment of up to £2 billion to accelerate AI innovation and research in the United Kingdom. In that release, AMD said it is working with Oriole Networks on ARIA’s Scaling Inference Lab, combining Oriole PRISM with AMD Instinct GPUs and AMD EPYC processors for what AMD described as expected to be the first large-scale AI system based on a pure photonic network.
Oriole Networks’ central claim is that photonic networking can help AI infrastructure move data with lower latency and reduced energy overhead compared with conventional electrical interconnect approaches. The company says its planned deployment is intended to support inference at scale, an area where data movement between accelerators, processors, and memory can become a major bottleneck.
The sources do not provide detailed benchmark results, deployment dates, power targets, or full system specifications. For that reason, the project is best understood as a planned deployment supported by company statements from Oriole and AMD, rather than an independently verified performance milestone.
The role of ARIA’s Scaling Inference Lab is significant because the program is focused on the practical challenges of running AI inference at larger scale. Inference workloads can be sensitive to latency and throughput, especially as organizations serve more users and larger models. Oriole’s announcement positions optical data movement as one way to reduce constraints that can limit data center performance.
AMD’s press release places the Oriole collaboration within a broader UK AI investment plan. The chipmaker said it is committing up to £2 billion to support AI innovation and research in the United Kingdom. As part of that announcement, AMD named Oriole Networks and ARIA’s Scaling Inference Lab, saying the effort will use AMD Instinct GPUs and AMD EPYC CPUs.
That confirmation matters because it corroborates the core hardware and partner details in Oriole’s own release. AMD did not publish independent test results for the Oriole system in the cited announcement, but it did confirm the collaboration and the intended combination of Oriole PRISM networking with AMD compute hardware.
AI data centers increasingly depend on fast communication between many accelerators. As models grow and inference demand rises, networking performance and energy use can directly affect cost and responsiveness. Photonic networking uses light rather than electrical signaling for data transmission, and companies in the field argue that optical approaches can improve bandwidth and efficiency for demanding AI workloads.
Oriole’s planned system is part of that broader push to rethink how AI hardware is connected inside data centers. The company’s claim that this will be the world’s first large-scale pure-photonic-network AI system should be treated as an attributed claim: both Oriole and AMD use similar language, but the provided sources do not include an independent technical audit.
The next important questions are practical ones: when the system will be fully operational, how it performs on real inference workloads, how much energy it saves, and whether the architecture can be deployed economically beyond a research setting. The cited announcements point to an ambitious infrastructure experiment, but they do not yet establish measured gains.
For now, the verified news is that Oriole Networks plans to deploy a photonic-network-based AI system for ARIA’s Scaling Inference Lab, and AMD has confirmed it is collaborating on the project with Instinct GPUs and EPYC CPUs. If the deployment delivers measurable improvements, it could add evidence for optical networking as a serious option for future AI data center design.
The announcement, published by Oriole Networks and distributed through EIN Presswire, describes the project as an effort to address latency, performance, and energy constraints in AI data centers.
Oriole says the system will use its PRISM technology alongside AMD hardware for ARIA’s Scaling Inference Lab.
AMD separately confirmed the collaboration in a press release about its commitment of up to £2 billion to accelerate AI innovation and research in the United Kingdom.
Continue reading
