Compute-in-memory chip market map lists 88 companies across 13 layers
The map shows capital flowing to big-SRAM and analog-adjacent designs rather than pure in-memory compute, a useful signal for anyone betting on inference-hardware roadmaps. Investors and hardware buyers can use the 13-layer breakdown, funding and revenue figures to compare vendors like d-Matrix, Axelera AI and Groq within one taxonomy. China-linked players appear across layers, including Houmo, Witmem and MemSilicon with RMB 400M+ raised, making it a reference point for tracking supply-chain alternatives.
A new market map, dated Nov 2026, catalogues 88 companies building compute-in-memory and near-memory chips across 13 technology layers. Valuations concentrate in neighbouring architectures: Cerebras sits at $56.4B and SambaNova at $11B in the big-SRAM layer, while d-Matrix holds a $2B valuation for digital in-memory data-center inference and Unconventional AI reaches $4.5B in the analog layer. The map covers RRAM, flash-based CIM, DRAM/HBM near-memory players such as Samsung HBM-PIM and SK hynix AiM, photonic computing (Lightmatter at $4.4B), neuromorphic chips like Intel Loihi, and research lines including TSMC's 254 TOPS/W 5nm digital macro. The full list and sources are published at cim-chip-map.vercel.app, with quarterly updates planned.
New market map: compute-in-memory chips in 2026 — 88 companies, 13 layers The biggest valuations aren't in true in-memory compute. Unconventional AI (analog, route undisclosed) is at $4.5B and d-Matrix (digital in-memory, data-center inference) at $2B, while Cerebras, the big-SRAM neighbor, sits at $56.4B. Layer by layer: 1. RRAM / memristor (8) — Mentium, Innostar (ReRAM pilot line), MemSilicon (RMB 400M+ raised). Weights stored as device conductance. 2. Flash / NOR (4) — Mythic ($125M raised), Witmem, UCUN, Anaflash. Mature flash cells doing analog multiply-accumulate. 3. Charge-domain & other analog (10) — Unconventional AI ($4.5B val.), EnCharge AI ($100M+ raised), Sagence (~$58M), Aspinity, Blumind. 4. Digital CIM at the edge (9) — Axelera AI ($250M+ raised), Houmo (160 TOPS), GSI Technology ($25.1M revenue), PIMCHIP. 5. Digital CIM for the data center (6) — d-Matrix ($2B val.), Fractile ($220M), Rain AI. In-memory plus chiplets for LLM inference. 6. DRAM / HBM near-memory (10) — XCENA ($570M val.), NeuroBlade ($83M), Samsung HBM-PIM (1.2 TFLOPS), SK hynix AiM, UPMEM. 7. Edge at-memory (3) — MemryX ($44M raised), Syntiant ($271.8M revenue in 2025), Untether AI (team joined AMD). 8. Big-SRAM chips (4) — Cerebras ($56.4B), SambaNova ($11B), Groq ($3.5B), Graphcore. Trade on-chip SRAM for bandwidth; not true in-memory. 9. Research lines (11) — imec (2,900 TOPS/W), TSMC (254 TOPS/W, 5nm digital macro), Intel 18A (147 TOPS/W), IBM NorthPole, Stanford NeuRRAM. 10. Photonic computing (7) — Lightmatter ($4.4B val.), Celestial AI ($2.5B), Neurophos ($110M), Q.ANT. Light for matrix math and interconnect. 11. Neuromorphic (7) — Intel Loihi / Hala Point (1.15B neurons), BrainChip, Innatera, SynSense. 12. Memory IP & devices (8) — Weebit Nano, Everspin ($20.6M quarterly revenue), Ferroelectric Memory Co (EUR 100M). 13. Equipment, materials & EDA (1) — Applied Materials. Data as of Nov 2026, public sources; company-reported numbers are labeled as such on the map. Bookmark this — we'll update it every quarter. Full list + sources: cim-chip-map.vercel.app 💬 2 🔄 0 ❤️ 2 👀 535 📊 2 ⚡
