ARIA will test Lumai's optical AI servers at rack scale in a commercial data center
Oxford spinout Lumai, led by Xianxin Guo, will put its Iris servers into ARIA's lab so researchers can test its claim of up to 90% power savings at rack scale.
By RuntimeWire Staff · Published · Updated
Primary source: PR Newswire Business Technology
Why it matters
Inference power is becoming a binding constraint on AI deployment, and Lumai is asking a neutral, government-backed lab to test whether light-based prefill can deliver large efficiency gains inside ordinary racks.
Dr. Xianxin Guo built Lumai around a physical switch: do the mathematics of AI inference with beams of light instead of electrons moving through transistors. On November 3rd, the Advanced Research and Invention Agency, or ARIA, announced that it will deploy Lumai's 3D optical inference technology in its Scaling Inference Lab, run by CommonAI CIC.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.