← Sticker Wall MODEL · DETAIL

Yixing Intelligence RISC-V AI Computing Super-Node Solution

1) Hardware sales: Charging intelligent computing centers for super-node hardware; 2) Solution licensing: Charging for c

MODEL

Key Fields

FIELD STAMPS

IndustryAI / LLM

RegionChina

ScaleMid-size

ChannelHybrid

📌 Background

As an open-source instruction set without licensing fees, RISC-V is accelerating its entry into AI inference and customized computing scenarios by 2026. According to reports from Zhidx.com, Yixing Intelligence unveiled the industry's first RISC-V AI computing super-node at WAIC 2026. Based on the self-developed Epoch chip and ELink high-speed interconnect, the solution achieves a single-rack aggregate bandwidth at the 100T level, supports full-interconnect scaling from 32 to 128 cards, and can scale horizontally to clusters of ten thousand cards. Its KernelFab tool reduces operator development time from weeks to days.

👤 Target Customers

Intelligent computing centers, cloud service providers, and enterprise large model inference clients.

💰 Revenue Streams

1) Hardware sales: Charging intelligent computing centers for super-node hardware; 2) Solution licensing: Charging for computing cluster solution licenses based on cluster size and project; 3) Acceleration services: Charging clients for customized inference acceleration services on a project basis; 4) Edge replication: Promoting and replicating to edge AI clients (opportunistic, revenue scale yet to be verified).

🧮 Cost Structure

Chip design and R&D, tape-out manufacturing costs, hardware-software co-development, and market promotion.

🛡️ Moat

Industry's first RISC-V AI computing super-node, hardware-software co-design capabilities, and first-mover advantage in the RISC-V ecosystem.

🔑 Keys to Success

Secure top-tier clients among intelligent computing centers and cloud service providers
Build a RISC-V inference software toolchain
Rapidly iterate on super-node interconnect and scheduling capabilities

⚠️ Risks

Standardization of RISC-V in the AI computing field may fall short of expectations
Fluctuations in customized demand from major clients impacting revenue
High costs associated with advanced process tape-outs

🏢 Cases

Yixing Intelligence releases the industry's first RISC-V AI computing super-node at WAIC 2026

📊 SWOT Analysis

Strengths

Open-source RISC-V architecture avoids patent licensing fees
Industry-first super-node solution offers scarcity
Hardware-software co-optimization enhances inference performance

Weaknesses

RISC-V software ecosystem is not yet mature
Gap exists compared to the NVIDIA CUDA ecosystem
Commercial validation period for the super-node solution is short

Opportunities

Driven by policies for domestic AI computing autonomy and control
Demand for customized computing driven by falling large model inference costs
Cloud service providers seeking secondary computing suppliers

Threats

Competitive pressure from NVIDIA and domestic GPU manufacturers
Risk of fragmentation in the RISC-V ecosystem
Capital dilution due to the wave of AI chip market entries

Sources