AI infrastructure startup Infinity has raised $15 million in funding at a valuation of $100 million. TechCrunch reported on July 20 that the round included Touring Capital, Principal VC, and researchers from OpenAI and Anthropic.
Infinity is developing software that helps AI chips run AI models more easily. Its core product is kernel software that can replace Nvidia CUDA.
Major AI development frameworks such as PyTorch and TensorFlow are built on CUDA, allowing developers to run applications written in languages such as Python on Nvidia chips without separate porting.
Many application startups, by contrast, lack the resources and technical expertise to write kernels, the low-level software that runs chips, or to move to other AI chips. Infinity is developing general-purpose kernel software that works across chips such as SRAM, GPUs, smartphone chips and systolic arrays. It aims to build a universal inference library that runs on all chips.
Infinity was founded last year by Jeremy Nixon (제러미 닉슨), a former Google Brain researcher. Nixon is also the founder of the hacker network community AGI House.
Chi-gyu Hwang
delight@d-today.co.kr