Etched Bets $300M on the Premise That Generic GPUs Can't Win Inference
The inference-chip startup raised at a $10.3B valuation, a wager that the AI stack is splitting into general-purpose training silicon and custom inference hardware.
Etched raised $300 million at a $10.3 billion valuation to keep building inference-specific silicon, according to @marcopapa99. We generally skip funding rounds, but the thesis underneath this one is worth isolating: that inference is a different problem than training, and that purpose-built chips will beat general-purpose GPUs on the economics that matter at scale.
Unlock the full briefing
Get every story in today's briefing, the full archive, and the daily AI intelligence brief.
All stories today
Full archive
Daily brief
Cancel anytime. Payments powered by Stripe.