Oxmiq Labs
Category: AI Chips / Semiconductors
Licensable GPU IP and AI infrastructure software company re-architecting the GPU stack from atoms to agents, enabling semiconductor companies and governments to build custom AI processors. Oxmiq Labs was founded in 2023. The company is led by Raja Koduri. Based in Campbell, California, United States. Team size: 11-50. Total funding raised: $60M. Latest round: Series A. Key investors include Samsung Catalyst Fund, Fundomo, MediaTek, Pegatron Venture Capital, CDIB-TEN, Darwin Ventures, Morgan Creek Digital, AM Intelligence Labs, Intel Capital.
- Founded
- 2023
- Headquarters
- Campbell, California, United States
- Team size
- 11-50
- Total funding
- $60M
Value proposition
Licensable GPU IP that allows companies and governments to build custom AI chips without developing from scratch — an "Arm for AI GPUs" — with full CUDA ecosystem compatibility, reducing AI compute costs by 50-100x.
Products and solutions
OxCore (licensable GPU architecture IP), OxQuilt (chiplet interconnect architecture), OxCapsule (dynamic GPU container technology), OxPython (CUDA-compatible runtime for non-NVIDIA hardware), AI Cloud platform (OxCapsule + OxPython + Agents)
Unique value
Only company offering a vertically integrated, licensable GPU IP stack (from silicon to cloud software) that is CUDA/PTX compatible and adapts to customer's existing memory, packaging, and foundry choices — enabling sovereign AI without vendor lock-in.
Target customer
Semiconductor companies, neocloud providers, AI infrastructure builders, governments pursuing sovereign AI programs, data center operators
Industries served
AI/Semiconductor, Sovereign AI infrastructure, Data center, Cloud computing, Edge AI
How they differentiate
Unlike NVIDIA (proprietary, non-licensable) or Tenstorrent (RISC-V CPU IP focused), Oxmiq offers GPU IP specifically designed for AI workloads with full CUDA ecosystem compatibility. Unlike custom silicon builders (Groq, Cerebras), Oxmiq licenses IP rather than selling finished chips, following the Arm licensing model for GPUs. The stack spans from renewable energy infrastructure to AI agents.
Main competitors
Tenstorrent (licensable AI silicon IP, RISC-V based), NVIDIA (dominant GPU ecosystem, non-licensable), Arm Holdings (licensable CPU IP, expanding into AI)
Key partnerships
AM Intelligence Labs (2 GW renewable-powered AI compute platform in Uttar Pradesh, India), MediaTek (strategic investor & partner), Pegatron (investor & manufacturing partner), SanDisk (memory-centric AI partnership)
Notable customers
20 companies and 10 universities using OxCore beta across ~300 GPUs (including ARM, AMD, Intel, Infineon, Global Foundries, Tenstorrent, Radisys as beta participants)
Major milestones
2023: Company founded by Raja Koduri, 2023-2025: Two-year stealth period developing GPU IP, Aug 2025: Emerged from stealth with $20M seed round led by MediaTek, Nov 2025: Launched public beta of software stack, Dec 2025: Released V2.2 with Linux support, Mar 2026: Partnered with AM Intelligence Labs for 2 GW renewable AI compute platform, Jul 2026: Raised $35M Series A co-led by Samsung Catalyst Fund and Fundomo ($60M total)
Growth metrics
~300 GPUs running OxCore beta; 20 companies + 10 universities using the stack; 9,159 LinkedIn followers; 47 employees on LinkedIn
Market positioning
Positioned as the "Arm for AI GPUs" — a licensable GPU IP alternative to NVIDIA's proprietary ecosystem, targeting sovereign AI programs and semiconductor companies that want to build custom AI accelerators without multi-billion-dollar R&D investments.
Geographic focus
India, Southeast Asia (Malaysia, Indonesia), US
Patents and IP
Team holds hundreds of patents collectively; specific patent portfolio not publicly enumerated
About Raja Koduri
Ex-Intel Executive Vice President & Chief Architect (IAGS); Ex-AMD Senior VP & Chief Architect, Radeon Technologies Group; Ex-Apple Director of GPU Architecture; Ex-ATI Technologies. M.Tech from IIT Kharagpur. Known for leading Intel Arc GPU development, AMD Vega/Navi architectures, and Apple's Retina display GPU transition.
Official website: https://oxmiq.ai