Tagged data center · show all articles
Qualcomm built an empire on mobile chips it doesn't fully control the ceiling of. Moving into AI datacenter inference is a bet on a market with no such ceiling — and a much more crowded field of competitors already established there.
Blackwell was a GPU with a name. Rubin is a whole platform — six distinct chips engineered to work as one system — and that shift says as much about where AI hardware is headed as any single spec.
Nvidia's market capitalization gets treated as a headline in itself. The revenue, backlog, and data-center figures underneath it are more informative than the number everyone quotes.
Individual accelerators keep getting faster. The copper connecting thousands of them together increasingly cannot keep up — which is why Nvidia is betting heavily on replacing that copper with light.
OpenAI's custom-accelerator agreement with Broadcom is larger than any other single deal in this cohort by power commitment. This briefing works through the numbers, the timeline, and what building at this scale actually requires.
Most accelerator deals get announced as a unit count or a dollar figure. AMD and OpenAI described theirs in gigawatts of power consumption — a framing that says as much about the industry's real constraint as it does about the chips themselves.