NVIDIA Watch posted an update
Nvidia’s AI-chip lead faces a challenge from custom processors, with Broadcom currently far larger than Marvell in AI chip revenue, according to The Motley Fool. The article puts Broadcom’s AI semiconductor revenue at $16.7 billion in its second quarter, compared with $2.1 billion in AI data-centre revenue for Marvell.
Why it mattersThe piece argues that Broadcom’s work with customers including Alphabet, Meta, Anthropic and OpenAI could help it gain ground as companies build chips for specific workloads. Its forecasts for Broadcom’s AI revenue are projections, not results in the bank. GPUs still have uses that custom chips do not, so this looks more like a contest over which workloads go where than a clean changing of the guard.
Discuss: Will custom AI chips take a meaningful share of workloads from Nvidia’s GPUs, or will the two approaches mostly serve different jobs?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.