Luminal, a San Francisco-based AI infrastructure startup, has raised $25 million in Series A funding in a round led by Standard Capital. Existing investors Y Combinator, Felicis, SagaVC, and Liquid also participated in the round.
With this investment, Luminal has raised a total of around $30.3 million in funding after it earned $5.3 million in seed funding in November 2025. The company has yet to reveal its valuation.
Luminal will use the ongoing funding to grow the engineering team, run successful go-to-market initiatives, and increase AI models as well as hardware platforms.
Founded by CEO Joe Fioti, along with co-founders Matthew Gunton and Jake Stevens, Luminal is developing technology that helps businesses run AI models faster and more efficiently across different computing hardware, including GPUs and application-specific integrated circuits (ASICs).
To run AI models effectively, it is necessary to have advanced computing hardware; optimizing such machines for different types of processors requires substantial engineering work. Due to these factors, working with different hardware suppliers or finding cheaper computing options is often a daunting task for software developers.
Luminal aims to solve this challenge through its search compiler and inference cloud. Its compiler automatically tests different ways of running AI models on specific hardware and selects the most efficient implementation. Its inference cloud allows developers to deploy and run AI models without managing the underlying computing infrastructure themselves.
The platform is suitable for helping with the high-volume AI applications such as the coding agents, voice agents, and image editing tools where quick processing and low-cost computing is quite important.
As stated by the company, the public cloud it operates processes around 90 trillion tokens a month and reached that point only within four months of its launch. Luminal has broadened its collaboration with Amazon and assigned a considerable number of its workloads to the company’s chips known as Trainium chips.
The company is also working with customers across different AI applications. For example, real-estate media company AutoHDR uses Luminal to process approximately 5 million images per month while another customer, Freebuff, uses its infrastructure for large-scale AI workloads.
The purpose of Luminal is to give developers the possibility of using AI models on various types of hardware instead of relying solely on a chip manufacturer. The funds recently raised will be used by the company for expanding its technology and providing infrastructure to accommodate the increasing demand for suitable AI computing.
