EVAS Intelligence’s RISC-V AI Chips Land 1.5 Billion Yuan Series B to Power Global Expansion

What makes the Epoch series different?

EVAS’s Epoch series is built on a RISC-V and RVV (RISC-V Vector) foundation and combines a TPU‑style architecture with the open instruction set. The chips natively support FP8 quantization, which has become a key format for modern AI workloads because it can balance compute efficiency and model accuracy. According to the company, this design aims to optimize three variables at once: performance per watt, deployment flexibility, and precision for large models.

On top of the chips, EVAS offers PCIe accelerator cards, complete servers, and large‑scale intelligent computing clusters. This product matrix gives it more control over end‑to‑end performance and lets customers buy into a stack that ranges from single cards to full data center deployments. For buyers, that reduces integration risk; for EVAS, it increases average deal size and defensibility.

How the new capital will be used

The company plans to use the 1.5 billion yuan round to scale production and commercialization of the current Epoch chips, invest in its next flagship product, and deepen its software‑hardware ecosystem. That includes tooling, frameworks, and platform software needed to make RISC-V infrastructure feel as turnkey as incumbent solutions for developers and enterprise teams.

Another priority is international expansion. EVAS wants to grow beyond domestic deployments and push its RISC-V AI computing offerings into global markets, positioning itself as an alternative for customers seeking more geographic and architectural diversity in their AI supply chains. If successful, that could create new pressure on established GPU providers and raise the profile of Chinese RISC-V players in the worldwide semiconductor ecosystem.

Why this matters for the AI hardware race

EVAS’s Series B comes amid broader momentum for AI hardware startups and RISC-V‑based designs. As model sizes grow and inference volumes spike, demand for specialized, cost‑efficient compute is expanding faster than leading suppliers can serve it. For investors, EVAS is a case study in how regionally backed, vertically integrated AI chip startups are trying to fill that gap with differentiated architectures and strong local support.

For founders building in adjacent spaces—whether infrastructure software, model deployment platforms, or application‑layer AI—moves like this are also a reminder. The AI stack is fragmenting into multiple hardware options, and go‑to‑market strategies increasingly need to account for a world where “Nvidia only” is no longer the default assumption.