zlacker

Give it time. AMD, AWS trainium/inferentia, and Google TPUs all compete here. The gap is mostly with software drivers/support.