# ARIA lab plans to test Lumai optical AI servers

> Lumai says its optical servers will be evaluated in a data-centre testbed.

By BIGCHANGE.AI Editorial

Published: 2026-11-03T19:03:39.147Z
Updated: 2026-11-03T19:03:39.147Z
Canonical: https://bigchange.ai/blog/aria-lumai-optical-ai-inference-test

![Conceptual illustration of a server cabinet and a beam of light.](https://bigchange.ai/api/media/file/lumai-server-light-concept.png)
AI-generated conceptual illustration by BIG CHANGE.

Lumai says its optical AI servers will be onboarded for evaluation at the UK’s ARIA-backed Scaling Inference Lab.

## The big change

- The proposed test would place Iris servers in a commercial data-centre environment.
- Lumai targets prefill, the stage that processes a model’s input before it answers.
- Rack-scale testing is planned; the company’s claimed power savings are not independently established.

[ARIA describes](https://aria.org.uk/opportunity-spaces/nature-inspired-computation/scaling-compute/scaling-inference-lab) the CommonAI-delivered lab as a testbed for rack-level systems. Partners contribute technology or resources, while ARIA funds integration and testing. The programme aims to compare complete systems with commercial alternatives.

[Lumai says](https://www.prnewswire.co.uk/news-releases/arias-scaling-inference-lab-deploys-lumais-optical-compute-technology-to-help-power-ai-era-302926852.html) Iris encodes numbers in light intensity and uses passive optical components for mathematical operations. Its proposed architecture targets matrix-heavy prefill while pairing optical processing with silicon for decoding. The company describes a hybrid system rather than a data centre operating entirely on light.

[NVIDIA’s general explanation](https://developer.nvidia.com/blog/demystifying-ai-inference-deployments-for-trillion-parameter-large-language-models/) separates inference into two phases: prefill processes input tokens and builds intermediate states used for the first output token; decoding generates subsequent tokens. The phases use resources differently, so an improvement in one stage alone does not establish an end-to-end gain across workloads.

In its November 3 announcement, Lumai says the approach could cut data-centre power consumption by up to 90%. It supplies no independent measurements, comparison conditions or specific workload for that claim.

## Sources

- [Lumai lab announcement](https://www.prnewswire.co.uk/news-releases/arias-scaling-inference-lab-deploys-lumais-optical-compute-technology-to-help-power-ai-era-302926852.html) — Company description of planned onboarding and optical inference.
- [ARIA Scaling Inference Lab](https://aria.org.uk/opportunity-spaces/nature-inspired-computation/scaling-compute/scaling-inference-lab) — Lab description of its general testbed and partner model.
- [NVIDIA inference deployment overview](https://developer.nvidia.com/blog/demystifying-ai-inference-deployments-for-trillion-parameter-large-language-models/) — Separate general explanation of prefill and decode.