MatX vs d-Matrix

MatX, founded by ex-Google TPU engineers, designs chips optimized specifically for large language models, with first silicon due in 2027.

d-Matrix builds in-memory compute chips optimized for efficient transformer inference at scale in data centers.