factual
Physics Limits Imply Much Faster Chip-to-Chip Latency
The theoretical floor for chip-to-chip latency is the speed of light, about two to three nanoseconds, versus about 4,000 nanoseconds on current NVIDIA products, which leaves large room for improvement in latency and power efficiency.
factualpending
Speaker
GavinEvidence Quote
“It's the speed of light. You can do it in just a handful — like two, three nanoseconds.”
Source
Etched - Building AI Hardware to Make Inference Faster and Cheaper - [Invest Like the Best, EP.480]— Invest Like the BestCreated: 11/17/2026, 7:08:01 AM
Your notes
Loading notes...