Develops a complete AI compute stack — custom silicon, software (compilers/OS) and datacenter-scale infrastructure. Delivers high-performance compute for large-scale AI training and inference workloads.