OXMIQ Labs is a GPU and AI architecture company with expertise spanning every layer of the AI stack, from renewable energy and data center infrastructure to silicon IP, electron-to-token systems (ETMsTM), and cloud software for AI factories and agents. OXMIQ develops licensable compute IP and adaptive AI infrastructure software. The company's mission is to re-architect the GPU stack: from Atoms to AgentsTM, making high-performance AI compute available, affordable, and within reach of design teams and builders who need it.
Watch Raja Koduri explain From Atoms to Agents in one minute →
The cost of converting electrons to tokens has to come down by 50x. That means rebuilding the whole stack. OXMIQ delivers licensable GPU IP and a vertically integrated compute stack that lets semiconductor companies and AI infrastructure builders own their compute, configure it for their workloads, and deploy it with full CUDA ecosystem compatibility.
Raja Koduri — Founder & CEO, OXMIQ Labs
License OxCore and OxQuilt to tape out a custom AI accelerator without re-architecting from scratch. CUDA and PTX compatible, and adaptive to the memory, packaging, and foundry choices you already have. Stop paying a tax to someone else's roadmap.
"We are very excited to co-lead OXMIQ's financing round and back Raja Koduri and the strong team at OXMIQ. OXMIQ's novel AI compute core and software platform enable heterogeneous, efficient custom inference compute solution for large-scale agentic workloads."
"Raja has built silicon at every layer of the stack, and he knows where the real constraints sit. Most compute IP makes the customer bend their memory, packaging, and foundry choices around the chip. OXMIQ IP adapts to what the customer already runs, turning supply chain constraints into leverage. We backed this team because they will define how compute gets built this decade."
"OXMIQ has an impressive, bold vision and a world-class team. The company's GPU IP and software innovations will drive a new era of compute flexibility across devices, from mobile to automotive to AI on the edge."
"OXMIQ has a unique technology and a great set of technologists working to shape the future of Memory-centric AI"
"While the industry sinks billions into CapEx, OXMIQ is building the IP architecture that compute runs on, far more capital-efficiently."
"OXMIQ gives AMI Labs access to some of the deepest hardware and systems expertise in the industry. Their team's experience across leading Silicon Valley companies is exactly what we need to architect infrastructure that can compete globally."
"AI deployments are increasingly chiplet-based, and the systems work behind those deployments is harder than the architecture work that precedes it. OXMIQ treats packaging and integration as first-class design inputs, which is how AI silicon actually gets built at scale. We invested because the team understands the path from compute IP to deployed systems, and we want to be part of that path."
"As AI inference scales across edge, cloud, and sovereign deployments, the demand for licensable, efficient GPU IP is accelerating. Raja and the OXMIQ team bring the rare combination of architectural vision and execution experience needed to build that platform. We've spent two decades investing across the semiconductor value chain in Asia and the US, and OXMIQ sits at the intersection of where we see the strongest long-term demand."
"It's a great honor to be working with the legendary Raja Koduri on OXMIQ Labs' breakthrough architecture with OxQuilt, OxCore and hardware agnostic software stack for agentic AI all the way from ultra-low power edge to datacenters."
"I am excited to join the OXMIQ board. Raja and this team are creating an open GPU architecture, a much needed step toward removing the artificial boundaries around AI innovation. As the industry concentrates around a few incumbents, this is more important than ever. OXMIQ's open, configurable foundation, which developers can build on and own, is exactly where compute should be heading."
Capital to accelerate product development and customer integration, including our $35M Series A.