AI infrastructure company Infinity has secured a $15 million funding round at a valuation of $100 million, backed by prominent investors including Touring Capital and researchers from OpenAI and Anthropic. The startup aims to develop universal software that can run on any type of AI chip, aiming to replicate the success of Nvidia’s CUDA software but for a wider range of devices.

Founded by Jeremy Nixon, who was previously at Google Brain and creator of AGI House, Infinity seeks to automate the complex process of writing low-level code. Their AI research agent Ignition tests, debugs, and optimises this code, making it easier for app developers to run their projects on various hardware without needing extensive technical knowledge.

The company's approach could significantly disrupt Nvidia’s market dominance by providing a more accessible solution for running AI models across different types of chips. Key customers include D-Matrix, an AI chip maker that competes with Nvidia. Infinity is also in talks with other major chip and cloud companies to expand its reach.

While humans remain integral for high-level direction, the agent handles much of the repetitive work, potentially reducing development times from months to mere days. The business model is based on performance gains rather than upfront licensing fees, making it an innovative approach in the AI infrastructure space.

In a nod to the future, Nixon envisions a world where automated systems can create and optimise their own code, mirroring his earlier success with Omega, which generated new machine learning algorithms. Infinity's vision is ambitious: a universal inference library that allows chips from SRAM to Systolic Arrays to operate at CUDA-like efficiency.