d-Matrix
Company
Chip Designers & Semis
In-memory compute for the economics of small-batch inference.
You're looking at a company record: what this company does in the AI stack, and every chip, technology, facility and company it's linked to below.
d-Matrix argues the inference market splits away from training hardware entirely: its Corsair cards compute inside SRAM to cut latency and power for interactive serving. Backed by Microsoft's M12 and Temasek.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Chips designed
Competes with
Related
Sources
Record reviewed as of 2026-01-15 — see Methodology.