Media Coverage
Startups to Watch - TechCrunch Disrupt
October 27, 2026
Lumai Editorial Team
Power is a practical limit on scaling AI. Lumai’s Xianxin Guo explains how the company approaches that problem with optical computing: using photons and interference to perform AI calculations...
What Happens in Prefill?
Every request to a large language model runs through two very different phases: prefill and decode. Most infrastructure teams spend their time optimizing decode, the token-by-token generation everyone can see in a streaming response. Prefill gets less attention and for a growing share of workloads this is a mistake....
.png)
The Matrix Math That's Breaking Your Inference Stack
Frontier AI is on track to need 1,000x more inference compute within five years, at a cost no one can build: $100 trillion and 1,000 GW. The bottleneck is matrix multiplication, and Lumai Iris solves it by performing that math with light instead of electricity, decoupling energy from compute and cutting inference energy 10x, no liquid cooling required.
The AI Infrastructure Bottleneck No One Is Solving
Every infrastructure roadmap for generative AI reads the same way: more GPUs, more high-bandwidth memory, faster interconnects, better token throughput. These investments matter, but specialization is only currently targeting one half of the inference pipeline. The other half, the stage where a model actually reads the prompt before it writes a word, still has no purpose-built hardware in production...
