AI Hardware Investment Tables
Private AI Hardware Investment Analysis
Risk-Adjusted Ranking for Non-VC Portfolio
January 2026
Executive Summary
Company Stage Valuation Total Raised Stack Position Risk Rating d-Matrix Series C $2.0B $450M AI Accelerator Medium FuriosaAI Series C+ $735M $246M AI Accelerator Medium Panmnesia Series A $250M $80M CXL/Memory High Axelera AI Series B ~$1.0B $200M+ Edge AI Medium SiMa.ai Series C $960M $355M Edge AI Medium-Low
1. d-Matrix (Recommended: Core Holding)
d-Matrix – Product & Competition Attribute Details Core Product Corsair AI inference platform using Digital In-Memory Compute (DIMC) chiplets Technology DIMC architecture eliminates memory wall; chiplet-based design with high-bandwidth BoW interconnects Stack Position Data center AI inference – targets LLM and transformer workloads at lower TCO than GPUs Real Competitors NVIDIA (H100/B100), Groq (LPU), Cerebras, AMD MI300, Intel Gaudi, SambaNova Differentiation Memory-bound compute efficiency; claims 3-5x better TCO for inference vs GPUs Key Risks NVIDIA ecosystem lock-in; software stack maturity; scaling production; customer adoption timing Exit Path IPO likely (2026-2028); strategic acquisition by hyperscaler or AMD/Intel possible
d-Matrix – Financial & Investment Attribute Details Valuation $2.0B (Nov 2025 Series C) Total Raised $450M across 3 rounds Latest Round $275M Series C (Nov 2025) – led by Temasek, Industry Ventures Key Investors Temasek, SK Hynix, M12 (Microsoft), Playground Global, Marvell Technology, EDBI Revenue Status Pre-revenue; commercial deployment began 2024; ramping customer pilots TAM AI inference market: $106B (2025) to $255B (2030); ASIC share growing to 40% Access Route EquityZen (secondary); direct allocation difficult at current stage
2. FuriosaAI (Recommended: Core Holding)
FuriosaAI – Product & Competition Attribute Details Core Product RNGD (Renegade) AI accelerator optimized for LLM inference in data centers Technology Custom NPU architecture; claims 2.25x better perf/watt vs GPUs for LLM inference Stack Position Data center AI inference – targets hyperscale and enterprise LLM deployments Real Competitors NVIDIA, AMD, Groq, Cerebras, Rebellions (Korean rival), Graphcore Differentiation Korean government backing; LG AI Research design win; OpenAI partnership demo Key Risks No blockbuster order yet; catching NVIDIA is massive challenge; Meta acquisition talks stalled Exit Path IPO (planning $300M pre-IPO round); strategic M&A (Meta was interested)
FuriosaAI – Financial & Investment Attribute Details Valuation $735M (~1T KRW) – Korean unicorn status (Jul 2025) Total Raised $246M across 4 rounds Latest Round $125M Series C Bridge (Jul 2025); planning $300M+ Series D Key Investors Korea Development Bank, Kakao Investment, Naver, DSC Investment, Industrial Bank of Korea Revenue Status Pre-revenue at scale; RNGD in production; LG AI Research customer TAM AI inference market: $106B (2025) to $255B (2030); data center AI chips to $457B by 2030 Access Route Limited – Korean VC/PE networks; potential pre-IPO allocation in Series D
3. Panmnesia (Recommended: High Conviction)
Panmnesia – Product & Competition Attribute Details Core Product CXL 3.1 switch chips and semiconductor IP for memory pooling/expansion Technology World's first CXL 3.1 IP with two-digit nanosecond latency; CXL-GPU for AI workloads Stack Position Memory interconnect layer – enables disaggregated memory architecture for data centers Real Competitors Astera Labs (public), Montage Technology, Microchip, Samsung (internal), Intel (internal) Differentiation Technology leader (back-to-back CES Innovation Awards); first CXL 3.1 silicon; KAIST pedigree Key Risks CXL adoption timing (2026-2027 inflection); hyperscaler internal development; early stage Exit Path Strategic acquisition by Samsung/SK Hynix/Broadcom; IPO possible if CXL market scales
Panmnesia – Financial & Investment Attribute Details Valuation $250M+ (Nov 2024 – largest Korean fabless Series A) Total Raised $80M (incl. govt grants) in ~2 years since founding Latest Round $60M Series A (Nov 2024) – led by InterVest Key Investors InterVest, Korea Investment Partners, KB Investment, Woori Venture Partners, Smilegate Investment Revenue Status Pre-revenue; IP licensing and silicon tape-out underway; strong hyperscaler interest TAM CXL memory: $2B (2025) to $20B+ (2030); CXL components $6B by 2034 Access Route Korean VC networks; potentially Series B in 2025-2026
4. Axelera AI (Recommended: Opportunistic)
Axelera AI – Product & Competition Attribute Details Core Product Metis AIPU (214 TOPS @ 15 TOPS/W); Europa next-gen (629 TOPS); Titania for data center Technology Digital In-Memory Compute (D-IMC) with RISC-V; targets memory wall problem Stack Position Edge AI inference – industrial, automotive, robotics, drones, surveillance Real Competitors Hailo, SiMa.ai, Qualcomm, NVIDIA Jetson, Intel Movidius, Google Edge TPU Differentiation European sovereignty play; Samsung backing; strong power efficiency; $100M+ pipeline Key Risks Edge market fragmentation; NVIDIA Jetson ecosystem; scaling from niche to mass market Exit Path Strategic M&A (Samsung?); IPO if data center expansion succeeds ($1-3B outcome likely)
Axelera AI – Financial & Investment Attribute Details Valuation ~$1.0B (implied from Jan 2025 credit facility) Total Raised $200M+ (equity + EU grants including €62M EuroHPC grant) Latest Round €62M EuroHPC grant (Mar 2025); seeking €150M+ new round (Aug 2025) Key Investors Samsung Catalyst Fund, Invest-NL, EIC Fund, CDP Venture Capital (Italy), imec.xpand, Bitfury Revenue Status ~$15M revenue (Aug 2025); $100M+ business pipeline; Metis in production TAM Edge AI chip market growing rapidly; AI semiconductor infrastructure $193B by 2027 Access Route European VC networks; potentially in upcoming €150M+ round
5. SiMa.ai (Recommended: Opportunistic)
SiMa.ai – Product & Competition Attribute Details Core Product MLSoC (Machine Learning System-on-Chip) for edge AI; software-centric platform Technology Full-pipeline ML SoC; no-code drag-and-drop development; 10x perf/power vs alternatives Stack Position Edge AI – industrial, automotive, retail, aerospace, defense, agriculture, healthcare Real Competitors Hailo, Axelera, Qualcomm, Ambarella, NVIDIA Jetson, Graphcore (edge) Differentiation Software-first approach; Cisco partnership for Industry 4.0; Michael Dell backing Key Risks Crowded edge market; customer concentration; scaling beyond niche verticals Exit Path IPO or strategic M&A by industrial/auto player; lower multiple exit ($1-3B likely)
SiMa.ai – Financial & Investment Attribute Details Valuation $960M (Jul 2025) Total Raised $355M across 7 rounds Latest Round $85M Series C (Aug 2025) – led by Maverick Capital Key Investors Maverick Capital, Fidelity, Point72, MSD Partners (Michael Dell), Lip-Bu Tan (angel) Revenue Status Revenue generating; commercial products shipping; Cisco partnership TAM Edge AI inference: ~$15B (2025) to $75B (2033); mobile/edge 120M+ units annually Access Route Best access among the group – later stage, multiple funding rounds available
Analysis based on publicly available information as of January 2026. This is not investment advice. Due diligence recommended before any investment decisions.
Discover more from Priory House
Subscribe to get the latest posts sent to your email.
Post navigation