The Rise of Domestic AI Chip Company Suoyuan Technology: Focusing on Ecosystems Rather than Pure Hardware Parameters – How Does Shanghai Help China Break Through in AI Computing Power?
Hello everyone, I'm your financial journalist. Today, we're talking about a star company that just made its debut on the STAR Market: Suoyuan Technology (688801.SH).
If you follow tech news, you might have heard of NVIDIA's dominant position in the AI chip industry. In China, there's been a long-standing question: Can China produce its own “NVIDIA”? Or at least, can we create domestically made AI chips that are competitive?
On September 11th, Suoyuan Technology went public, marking not only a significant milestone for the company but also a signal that China's AI chip industry is moving from a phase of simply competing on hardware parameters to one where ecosystems, product definition, and collaboration are key factors.
To make this complex topic easier to understand, I've broken down the news into five key points and explained the underlying logic in plain language.
---
1. The Big Picture: Focus on Defining Products, Not Just Chips
Many people think that chip companies just design chips, manufacture them, and sell them. However, Zhao Lidong, the chairman of Suoyuan Technology, made a poignant point: “Following established paths essentially means repeating what others have done, making it difficult to build original technological advantages.”
- The old approach (engineering replication): If others use NVIDIA’s architecture, you just copy it. While it might work, you’ll always be lagging behind and vulnerable to supply chain issues.
- The new approach (product definition): Instead of copying others, we consider what customers will need in the next three years. For example, if a customer plans to use video generation next year and AI agents the year after, we design chips with the necessary capabilities.
To put it simply:
It’s like making phones. Engineering replication would be making a phone with a similar screen size and camera resolution to the iPhone. Product definition means optimizing the phone for specific user needs, such as better night mode photography or longer battery life.
Suoyuan believes that those who can define products hold the true value. This isn’t about copying; it’s about setting the standards.
---
2. Technological Path: Choosing a Different Route
In the AI chip world, there’s a well-known software ecosystem called CUDA (used by NVIDIA). Over 90% of AI developers worldwide use it because it’s like a “universal key” that can open many doors. Most domestic chip manufacturers choose to be compatible with CUDA or mimic NVIDIA’s GPGPU (general-purpose graphics processing units) architecture to survive. But this means following NVIDIA’s rules, making it hard to stand out.
Suoyuan, however, has taken a different, more “core” approach:
- DSA (Domain-Specific Architecture): They have chosen this path and intentionally do not support the CUDA ecosystem.
- Why this?
- Autonomy and control: By not being compatible, Suoyuan can control everything from the instruction set to the drivers, building a strong competitive advantage.
- Cost-effectiveness: DSA is optimized for specific AI tasks (e.g., large language models, image recognition), making it more efficient and cost-effective in these areas.
The results so far?
Suoyuan’s third-generation product, the S60, has shipped 160,000 units and supports over 300 use cases. This shows that a “domain-specific” approach works well in the AI boom, and its lower cost makes it popular with large internet companies.
---
3. Business Logic: Open Source Models Lower the Bar
Previously, domestic chips struggled to sell because of the lack of a strong software ecosystem (like CUDA). But things are changing.
Zhao Lidong predicts that 2025 will be a turning point for China’s AI chip industry due to the rise of open-source models:
- Previously: Training a large model with billions of parameters required expensive NVIDIA GPUs. Now, open-source models like Llama and Qwen have lowered the barriers.
- No need for huge budgets: You can use open-source models to train smaller, more affordable models or build on existing ones.
- Greater accessibility: This opens up opportunities for startups and small institutions to participate in AI development.
What does this mean for domestic chips?
- Lower entry barriers: Top-tier computing power is no longer just for giants; startups and small organizations can use affordable solutions.
- Cost-effectiveness matters: With Tencent’s support, Suoyuan’s chips are highly competitive in terms of cost-performance. This has led to increased demand.
Data supports this: Suoyuan’s revenue grew from 300 million in 2023 to 990 million in 2025, with a 279% increase in the first half of 2026, indicating strong market demand.
---
4. The Power of Shanghai
Many wonder why Suoyuan is based in Shanghai. Zhao Lidong explains that Shanghai has a complete ecosystem for chip design, manufacturing, packaging, testing, and end applications:
- Talent pool: Shanghai attracts 40% of China’s integrated circuit professionals.
- Collaborative advantages: Close proximity of chip design, packaging, and server manufacturing companies reduces innovation costs and enhances efficiency.
- Government support: From providing subsidies to building ecosystems, Shanghai helps companies from research to mass production.
In simple terms:
In other cities, you might have to travel far for different services. In Shanghai, everything is within easy reach. Shanghai’s integrated circuit industry is worth 580 billion yuan, making it the fourth-largest in the world and the largest in China.
---
5. Future Directions: Moving Beyond Single Breakthroughs to a Comprehensive Ecosystem
Suoyuan’s listing is just the beginning. Zhao Lidong talks about “core-model collaboration”:
- What is it? It’s the combination of domestic chips, models, and software platforms.
- Why is this important? Each component must work together effectively. A strong chip without a compatible model is useless, and a strong model without a suitable chip is as well.
Suoyuan’s next steps include:
- Deepening partnerships with top Shanghai companies in CPU, storage, networking, and modeling.
- Integrated solutions: Providing both hardware and software optimization services.
- Building a closed loop: From research to application, achieving full autonomy and control.
In summary, Suoyuan Technology’s listing marks a new phase in China’s AI chip industry, one where ecosystems are the focus. Technologically, China is moving away from blindly following NVIDIA and towards autonomy with DSA. Commercially, it’s leveraging open-source models to offer cost-effective solutions for the mid-range market. Geographically, Shanghai’s ecosystem provides the perfect foundation for collaboration. Strategically, the company is shifting from selling chips to providing comprehensive computing solutions.
For us as consumers, this means that the computing power behind AI services in China will increasingly come from domestically made chips like Suoyuan’s. This is not just about chip companies’ stock prices but also about the security and independence of our national digital infrastructure.
In one sentence: China’s breakthrough in AI chips is about building a more complete ecosystem, not just creating better components.