AI infrastructure startup Infinity has raised US$15 million in funding at a US$100 million valuation to develop software that enables artificial intelligence models to run efficiently across different types of AI chips.
The funding round included investors such as Touring Capital, Principal VC and researchers from OpenAI and Anthropic.
Building a CUDA alternative
Founded in 2025, Infinity is developing a software stack designed to work across multiple chip architectures, including GPUs, SRAM-based processors, mobile chips and systolic arrays.
The company aims to provide an alternative to Nvidia’s CUDA platform, which has become the dominant software layer for AI development.
Founder Jeremy Nixon, a former Google Brain researcher, said Infinity’s goal is to automate the creation and optimization of low-level software used to run AI models.
AI agent optimizes chip performance
Infinity’s AI research agent, called Ignition, automatically writes, tests and optimizes inference code for different processors.
According to the company, the system continually improves performance by rewriting code based on testing results, reducing work that previously required months of engineering to just days or hours.
Infinity said customers include AI chip developer D-Matrix, while discussions are underway with other semiconductor and cloud companies.
Rather than charging software licensing fees, the startup generates revenue by sharing in the performance improvements and cost savings delivered to customers.
Infinity currently employs 26 people across engineering, operations and product development.