Software, there are 100s of ai hardware startups. It’s easier relatively to build a domain specific chip, but building a compiler stack that integrates with a m

Software, there are 100s of ai hardware startups. It’s easier relatively to build a domain specific chip, but building a compiler stack that integrates with a major framework and gets good perf from your chip is incredibly difficult. Getting people to switch from nvidia? Even harder