The future of software is fabless

Early-stage startups have always been bottlenecked by their ability to ship more product. Raise some money > hire engineers > ship something > raise and hire more, etc.

LLMs are changing the way 0 to 1 works. Founders would be happy to spend $100k in tokens a week to get a product to market quickly with a very small team. We are already seeing many startups with more $Ms in ARR than employees. Early stage VC will barbell to be either small rounds to fund coding agents inference or very large rounds to aggressively scale startups once the product hits initial PMF.

At Kernel Labs we want to build the TSMC of startups. We believe that startups are context arbitrages, and what founders will need in the future is not just capital, but the infrastructure to more effectively turn their unique context into their product. Just like NVIDIA doesn't need to build fabs, we shouldn't expect every founder to rebuild a coding factory from scratch.

If this sounds interesting and you want to help build it, shoot an email to alessio at this domain.