OXMIQ Secures $35 Million to Drive AI Silicon Innovation
July 2, 2026, 10:03 am
Location: United States, North Carolina, Chapel Hill
Employees: 11-50
Founded date: 2004
OXMIQ, founded by Raja Koduri, secured $35 million in Series A funding. Total capital reached $60 million. The company pioneers a licensable GPU and AI architecture. Its OxCore technology enables custom AI silicon for diverse applications. This investment scales OxCore development. It aims to lower AI computing costs. OXMIQ’s innovative IP-first model reshapes semiconductor manufacturing. It offers flexible, energy-efficient solutions for the AI era.
OXMIQ, a leader in AI architecture, announced significant new funding. The company closed a $35 million Series A round. This capital injection raises its total funding to $60 million. Fundomo and Samsung Catalyst Fund co-led the investment. Other participants included MediaTek, AM Intelligence Labs, Pegatron Venture Capital, CDIB-TEN, Darwin Ventures, and Morgan Creek Digital. This funding fuels OXMIQ's mission. It supports the scaling of its OxCore architecture.
Raja Koduri founded OXMIQ. He is a recognized expert in GPU design. His vision is bold. OXMIQ aims to re-architect the entire GPU stack. This spans from "Atoms to Agents." The company builds silicon IP, configurable systems, and software platforms. These tools lower the cost of intelligence across the stack. OXMIQ makes AI more accessible. It empowers more builders.
At the core of OXMIQ's innovation lies OxCore. This is a scalable, licensable GPU architecture. It allows semiconductor companies to create custom AI silicon. AI system builders also benefit. They avoid full chip design programs. This accelerates development. It reduces entry barriers.
OxCore integrates three critical compute engines. A CUDA-compatible GPU engine processes general-purpose tasks. A tensor processing engine handles AI-specific workloads efficiently. An orchestration engine coordinates workloads. It manages agents across the system. This tight integration is unique. Functions typically split across multiple chips are unified. This design reduces data movement. It significantly improves energy efficiency for demanding AI tasks.
The architecture emphasizes near-memory compute. This minimizes latency. It maximizes throughput. OxCore scales broadly. It supports single-core AI deployments. It also powers large-scale data center configurations. The technology is currently running on FPGA. Live demonstrations showcase its capabilities.
OXMIQ's innovation extends beyond OxCore. The company developed OxQuilt. This is a chiplet integration architecture. It combines heterogeneous compute chiplets. It integrates memory into a single package. Many AI silicon designs are tied to specific foundries. They often depend on particular memory types. OxQuilt offers flexibility. It adapts across diverse supply chains.
OxQuilt provides robust configuration tools. Customers can design across logic process nodes. They can select various memory types. Different interconnect standards are supported. Advanced packaging options are available. This adaptability is crucial. It empowers designers. They create optimized solutions.
A comprehensive software stack complements OXMIQ's hardware. OxCapsule offers high-level orchestration. It also provides low-level kernel optimization. This ensures peak performance. OxPython is another key component. It enables existing CUDA and PyTorch code to run on OxCore. No code changes are required. This ensures portability. Developers maintain their existing workflows. OxPython has been validated on third-party platforms. Live demonstrations are available.
OXMIQ operates on an IP-first business model. This strategy prioritizes capital efficiency. The company focuses on new architecture IP development. It avoids full system-on-chip (SoC) development costs. This model generates revenue from customer engagements. It preserves capital for continued stack development. This approach is agile. It fosters sustainable growth.
The company's board and advisory team have expanded. Jim Keller joined the board of directors. He is CEO of Tenstorrent and a renowned chip architect. Dr. Ker Zhang remains an existing board member. Dr. Valluri “Bob” Rao joined as an advisor. He is a former Intel process technology Fellow. These additions bring extensive industry experience. They strengthen OXMIQ's leadership.
OXMIQ targets a wide range of partners. Semiconductor companies are key collaborators. Neocloud providers benefit from its technology. AI system builders find custom solutions. Physical AI and robotics companies also engage. These partners seek control over their compute roadmaps. OXMIQ provides the tools for that control.
The demand for AI infrastructure is soaring. Token demand outpaces current capabilities. OXMIQ directly addresses this challenge. Its open, configurable foundation democratizes AI innovation. It lowers the cost barrier. It widens access to advanced AI tools. This fosters broader participation. It unlocks new possibilities in AI development.
OXMIQ’s technology promises significant impact. It can redefine how AI compute is built. It offers energy-efficient solutions. It promotes custom silicon development. The company is headquartered in Campbell, California. A development site operates in Hyderabad, India. OXMIQ is poised for leadership. It will shape the future of AI.
OXMIQ, a leader in AI architecture, announced significant new funding. The company closed a $35 million Series A round. This capital injection raises its total funding to $60 million. Fundomo and Samsung Catalyst Fund co-led the investment. Other participants included MediaTek, AM Intelligence Labs, Pegatron Venture Capital, CDIB-TEN, Darwin Ventures, and Morgan Creek Digital. This funding fuels OXMIQ's mission. It supports the scaling of its OxCore architecture.
Raja Koduri founded OXMIQ. He is a recognized expert in GPU design. His vision is bold. OXMIQ aims to re-architect the entire GPU stack. This spans from "Atoms to Agents." The company builds silicon IP, configurable systems, and software platforms. These tools lower the cost of intelligence across the stack. OXMIQ makes AI more accessible. It empowers more builders.
At the core of OXMIQ's innovation lies OxCore. This is a scalable, licensable GPU architecture. It allows semiconductor companies to create custom AI silicon. AI system builders also benefit. They avoid full chip design programs. This accelerates development. It reduces entry barriers.
OxCore integrates three critical compute engines. A CUDA-compatible GPU engine processes general-purpose tasks. A tensor processing engine handles AI-specific workloads efficiently. An orchestration engine coordinates workloads. It manages agents across the system. This tight integration is unique. Functions typically split across multiple chips are unified. This design reduces data movement. It significantly improves energy efficiency for demanding AI tasks.
The architecture emphasizes near-memory compute. This minimizes latency. It maximizes throughput. OxCore scales broadly. It supports single-core AI deployments. It also powers large-scale data center configurations. The technology is currently running on FPGA. Live demonstrations showcase its capabilities.
OXMIQ's innovation extends beyond OxCore. The company developed OxQuilt. This is a chiplet integration architecture. It combines heterogeneous compute chiplets. It integrates memory into a single package. Many AI silicon designs are tied to specific foundries. They often depend on particular memory types. OxQuilt offers flexibility. It adapts across diverse supply chains.
OxQuilt provides robust configuration tools. Customers can design across logic process nodes. They can select various memory types. Different interconnect standards are supported. Advanced packaging options are available. This adaptability is crucial. It empowers designers. They create optimized solutions.
A comprehensive software stack complements OXMIQ's hardware. OxCapsule offers high-level orchestration. It also provides low-level kernel optimization. This ensures peak performance. OxPython is another key component. It enables existing CUDA and PyTorch code to run on OxCore. No code changes are required. This ensures portability. Developers maintain their existing workflows. OxPython has been validated on third-party platforms. Live demonstrations are available.
OXMIQ operates on an IP-first business model. This strategy prioritizes capital efficiency. The company focuses on new architecture IP development. It avoids full system-on-chip (SoC) development costs. This model generates revenue from customer engagements. It preserves capital for continued stack development. This approach is agile. It fosters sustainable growth.
The company's board and advisory team have expanded. Jim Keller joined the board of directors. He is CEO of Tenstorrent and a renowned chip architect. Dr. Ker Zhang remains an existing board member. Dr. Valluri “Bob” Rao joined as an advisor. He is a former Intel process technology Fellow. These additions bring extensive industry experience. They strengthen OXMIQ's leadership.
OXMIQ targets a wide range of partners. Semiconductor companies are key collaborators. Neocloud providers benefit from its technology. AI system builders find custom solutions. Physical AI and robotics companies also engage. These partners seek control over their compute roadmaps. OXMIQ provides the tools for that control.
The demand for AI infrastructure is soaring. Token demand outpaces current capabilities. OXMIQ directly addresses this challenge. Its open, configurable foundation democratizes AI innovation. It lowers the cost barrier. It widens access to advanced AI tools. This fosters broader participation. It unlocks new possibilities in AI development.
OXMIQ’s technology promises significant impact. It can redefine how AI compute is built. It offers energy-efficient solutions. It promotes custom silicon development. The company is headquartered in Campbell, California. A development site operates in Hyderabad, India. OXMIQ is poised for leadership. It will shape the future of AI.