Progress in AI is moving from training bigger models to letting them think longer on the problem in front of them. That makes inference the workload that matters most, and it deserves hardware designed for it from the silicon up.
That’s what we’re building.