EVAS Intelligence, a Chinese developer of RISC-V-based AI processors, has raised nearly 2 billion yuan in a new financing round as it tries to move its architecture from individual accelerators into large-scale cloud computing systems. TechTimes, in reporting published September 20, converted the financing to approximately $295 million, while several financial outlets have described the same yuan amount as roughly $280 million. The company’s post-money valuation was estimated at nearly 15 billion yuan, or a little above $2 billion depending on exchange rates.

More than 20 institutions participated, according to Chinese financial reporting. They included Huatai Innovation, Eastern Bell Capital, Zhongxin Juyuan, Andon Health, Tongfu Microelectronics and a collection of other financial and semiconductor-sector investors. Public reports did not provide a complete breakdown of how much each investor committed.

The financing gives EVAS additional resources to commercialize its Epoch family of cloud AI chips and develop its next generation of processors. Chinese reports say the first Epoch processor has reached mass production and customer delivery, while a newer cloud-computing chip has completed tape-out, the stage at which a finished design is sent toward fabrication.

EVAS was founded in 2022 and has increasingly positioned itself as a provider of an alternative AI computing stack rather than simply another accelerator designer. Its approach combines the RISC-V instruction set with the company’s own EVAMIND acceleration architecture. Earlier this year, EVAS also raised 1.5 billion yuan in a Series B financing round intended to support Epoch production, research and development, software development and broader commercialization.

At the World Artificial Intelligence Conference in Shanghai this summer, EVAS introduced what it described as a RISC-V AI “SuperNode,” a rack-scale computing system combining Epoch processors with the company’s ELink interconnect and liquid-cooled infrastructure. The announcement reflected a broader shift in AI hardware competition: chipmakers increasingly have to provide networking, systems software and cluster architecture alongside raw processor performance. EVAS has made ambitious scaling claims for the system, but independently verified benchmarks comparing its platform with established Nvidia, AMD or Huawei systems remain limited.

The RISC-V component is what gives the financing unusual relevance in the United States. RISC-V is an open instruction-set standard that companies can implement without obtaining the type of proprietary architectural license associated with some competing processor ecosystems. That gives Chinese chip developers a foundation on which they can design processors without relying on a foreign company for access to the basic instruction set.

That distinction has previously drawn attention in Washington. A bipartisan group of U.S. lawmakers wrote to the Commerce Department in 2023 asking it to examine whether American participation in RISC-V development with Chinese entities should face additional restrictions. The lawmakers argued that China could use open processor technology to reduce the effectiveness of U.S. semiconductor controls. That was a policy proposal and expression of concern, not a determination that RISC-V itself violated U.S. law.

RISC-V International, the nonprofit organization that maintains the standard, has taken a different position on the implications of its governance structure. The organization says it incorporated in Switzerland to preserve long-term international collaboration and reduce uncertainty, and it explicitly states that the move did not circumvent existing restrictions. It also notes that publicly available RISC-V intellectual property has historically not been subject to export controls simply because of the architecture itself.

That means describing an EVAS processor as inherently outside U.S. export controls would be too broad. U.S. semiconductor restrictions are not defined solely by which instruction set a chip uses. Bureau of Industry and Security rules can impose licensing requirements based on the performance of advanced computing products, the origin of controlled technology, semiconductor manufacturing equipment, specified end uses and end users, and foreign-produced items with certain connections to U.S. technology.

The Commerce Department has continued adjusting those rules. In January 2026, BIS revised its policy for exports of Nvidia H200, AMD MI325X and similar processors to China, allowing license applications to be reviewed case by case when specified security and compliance requirements are met. Earlier measures also tightened controls on semiconductor manufacturing equipment and high-bandwidth memory used in advanced AI systems.

For EVAS, RISC-V therefore addresses one layer of technological dependency rather than every potential constraint in a modern AI-chip supply chain. A commercial accelerator still requires manufacturing capacity, packaging, memory, design tools, networking components and software. The regulatory treatment of those inputs depends on their origin, specifications, suppliers and end users, not simply on whether the finished processor executes RISC-V instructions.

The larger competitive question is whether EVAS can turn architectural independence into a usable AI ecosystem. Nvidia’s position in artificial intelligence has been reinforced not only by GPU performance but by CUDA, libraries, compilers, networking technology and years of developer adoption. Chinese alternatives face the same systems-level challenge even when the underlying chip architecture is available without proprietary licensing.

EVAS is attempting to answer that challenge with a vertically integrated strategy spanning processors, interconnects, computing modules, rack-scale systems and software. Its latest financing suggests investors are prepared to fund that effort at a substantially larger scale. The next test will be commercial: whether production deployments can demonstrate competitive performance, reliability and software compatibility across large AI workloads.

For U.S. semiconductor companies and policymakers, the round also illustrates an important consequence of technology restrictions: pressure on access to established products can accelerate investment in alternative architectures and domestic supply chains. RISC-V gives Chinese developers one such architectural path. Whether EVAS can turn that path into a globally competitive cloud AI platform remains unproven, but nearly 2 billion yuan of new capital gives the company considerably more room to try.