Luminal raises $25M Series A led by Standard Capital for AI inference

Luminal is a San Francisco company building a compiler and inference service that searches for efficient ways to run AI models on the hardware a customer has available. Standard Capital led the Series A, with Y Combinator, Felicis, Saga Ventures and Liquid 2 Ventures participating. A competing report put the round at $23 million, but this summary uses the $25 million figure from Axios, which the company has not confirmed in a primary announcement. No valuation has been disclosed.

Inference costs recur every time a customer uses a deployed product, so per-request efficiency now shapes margins for AI application companies. Luminal must compete with optimizations from chip makers, cloud providers and other inference vendors. Its performance claims still need testing across workload sizes, hardware types and latency targets.

For founders, the round shows that investors will fund software layers that reduce dependence on one GPU supplier. Teams building AI products should track how inference costs change as more hardware options become viable.