OXMIQ Labs has raised $35 million in a Series A funding round to expand its licensable GPU architecture as demand grows for customized artificial intelligence (AI) chips that can be developed without designing full system-on-chip (SoC) platforms.
The financing brings the company’s total funding to $60 million. The round was co-led by Fundomo and Samsung Catalyst Fund, with participation from MediaTek, Pegatron Venture Capital, AM Intelligence Labs, CDIB-TEN, Darwin Ventures, Morgan Creek Digital and Intel Capital, among other investors.
Founded by former Intel and AMD graphics executive Raja Koduri, OXMIQ is developing GPU intellectual property (IP) that semiconductor companies and AI infrastructure providers can license to build customized AI processors.
The company said the funding will be used to scale OxCore, its configurable GPU architecture, which combines graphics processing, tensor computing and CPU orchestration into a single licensable compute core. The architecture is designed to reduce development costs for companies building AI accelerators while supporting deployments ranging from edge devices to hyperscale data centers.
AI chip developers are increasingly exploring modular approaches as demand for AI inference continues to rise. Rather than developing complete chips from the ground up, companies are looking to license reusable IP and combine it with chiplet-based designs to shorten development cycles and reduce costs.
OXMIQ said its OxQuilt chiplet architecture enables customers to combine different compute chiplets and memory technologies within a single package. Unlike conventional AI processors that are often tied to specific foundries or memory technologies, the platform is designed to support multiple manufacturing processes, packaging technologies and interconnect standards.
The company also offers a software platform that allows existing CUDA and PyTorch applications to run on OxCore without requiring code modifications. The software stack is intended to simplify migration for AI developers while supporting new AI models as they become available.
OXMIQ said demonstrations of both its hardware and software platforms are currently running on field-programmable gate arrays (FPGAs).
The funding comes as AI infrastructure spending accelerates globally. Demand for AI accelerators has surged alongside the rapid adoption of generative AI, prompting semiconductor companies and cloud providers to invest in alternative GPU architectures and custom silicon to reduce dependence on a small number of incumbent suppliers.
Alongside the financing, OXMIQ expanded its leadership team. Jim Keller, chief executive of Tenstorrent and a veteran semiconductor architect, has joined the company’s board of directors. Former Intel Fellow Bob Rao has also joined as an advisor.
Several strategic investors said the funding reflects growing interest in more flexible AI computing architectures. MediaTek, an existing investor, participated in the round, while Samsung Catalyst Fund and other investors cited increasing demand for customizable AI infrastructure across cloud, edge and enterprise applications.
The company said it is working with semiconductor companies, AI system developers, cloud infrastructure providers and robotics companies interested in licensing its technology for future AI chip development.
