Infinity raises $15M to build CUDA alternative
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Infinity raised $15 million at a $100 million valuation from Touring Capital, Principal VC, and researchers at OpenAI and Anthropic.
- The startup is building CUDA-alternative kernel software that works with any chip architecture — SRAM, GPUs, phone chips, and Systolic Arrays — rather than locking AI workloads to Nvidia's stack.
- Its AI research agent Ignition writes, tests, debugs, and auto-rewrites the low-level inference code, and is self-optimizing across different proprietary chip designs.
- AI chip maker D-Matrix is already a customer, and Infinity is in talks with other major chip and cloud companies, founder Jeremy Nixon told TechCrunch.
- Jeremy Nixon, a former Google Brain researcher and creator of the AGI House hacker community, started Infinity last year to pursue what he calls "automated invention."
- Infinity charges a cut of performance gains and cost savings (measured in tokens per second) rather than an upfront license fee, and currently has 26 employees.
Why it matters: Infinity's Ignition agent targets Nvidia's CUDA lock-in — the software moat that keeps AI workloads running on Nvidia hardware. Customer D-Matrix and other chip challengers stand to port inference workloads in hours or days, per the company, versus the years or months of manual kernel engineering the article describes.
