Hello! I'm your financial analysis assistant. This news article about the "Four Little Dragons" of domestic GPUs—Suoyuan Technology, Moore Threads, Muxi Shares, and Beren Technology—is incredibly informative and full of the intricacies of capital market strategies and industrial logic. To help you understand this complex news easily, I've broken it down into a "Core Summary" and **"Five In-Depth Analyses." We'll try to use plain language to explain the underlying business logic.
---
【Core Summary】
In one sentence:
Although the four domestic GPU companies have all gone public with a combined market value of over 600 billion yuan, they haven't yet successfully turned a profit. They are currently in a phase of "burning money to gain market share" and face significant pressure from restrictions on their shares and competition from giants like Cambricon and Huawei Ascend. The market is shifting from focusing on the concept of having domestic chips to assessing which companies can truly survive and make a profit.
Key statistics:
- Market value: Approximately 613.2 billion yuan in total.
- Revenue: All saw rapid growth in the first half of the year, but after deducting non-recurring expenses, they all incurred losses.
- Stock prices: After a sharp rise at the time of listing, Moore Threads' stock price has fallen by more than 60%, and Muxi's by more than 50%, indicating that some of the initial bubble has burst.
- Competitive landscape: Huawei Ascend holds about 50%-60% of the market share, followed by Cambricon, with the Four Little Dragons competing for the remaining niche market.
---
【In-Depth Analyses】
1. Four Different Approaches: No One-size-Fits-All Solution
Although these companies are all producing domestic GPUs, they each follow distinct business models with their own advantages and disadvantages:
- Suoyuan Technology (tightly tied to Tencent):
- Model: Relying on a major partner. Tencent is both the largest shareholder and the main customer.
- Current situation: Revenue has grown rapidly (279%), but the gross margin is low (around 30%) due to customizations for Tencent, limiting bargaining power.
- Strengths and weaknesses: It holds 6.6 billion yuan in advance payments from Tencent (a contractual liability), indicating stable orders. However, this is a liability and not easily transferable to other clients. In other words, it relies on Tencent for revenue but is constrained by its relationship with them.
- Muxi Shares (diversified customer base):
- Model: Seeking a wider range of customers to reduce reliance on a single giant.
- Current situation: It has the highest gross margin (57%), indicating profitable products. It even made a profit after deducting non-recurring expenses in the second quarter.
- Strengths and weaknesses: Although it made a profit in one quarter, it still lost money for the year. There's also a performance-based agreement that could extend the CEO's lock-up period if profits don't improve in the next few years. Its products are competitive, but the company is not yet large enough to generate stable profits.
- Moore Threads (comprehensive approach):
- Model: Trying to be all-in-one, producing both chips and system clusters, with the largest revenue.
- Current situation: It has the highest revenue, but its gross margin is declining.
- Strengths and weaknesses: It has taken on many low-margin system integration projects, which increases revenue but strains profitability. This means it's spreading resources between different areas, potentially affecting research and development.
- Beren Technology (focusing on cloud computing):
- Model: Specializing in high-demand cloud training chips with high technical barriers.
- Current situation: Revenue growth is the fastest (1997%), but the base is small (only tens of millions last year).
- Strengths and weaknesses: Its gross margin has improved the most, but it still lags behind competitors. It faces challenges with HBM (high-bandwidth memory) supply, increasing costs.
2. The Harsh Reality: The Market Is Already Divided
Many think domestic substitution is about starting from scratch, but in reality, the "from 0 to 1" phase is over, and now it's about competing for a larger share of the market:
- Market landscape: NVIDIA's share in China's AI chip market has dropped from 95% to 8%.
- Huawei Ascend and Cambricon: They hold about 50%-60% of the domestic market, with the Four Little Dragons competing for the rest.
- Conclusion: The remaining space is contested by companies like AMD, Pingtouge, and Kunlun Core. This is no longer a blue ocean market but a red ocean.
- Comparison with Cambricon: Cambricon had revenue of 6 billion yuan and a net profit of 2.3 billion yuan in the first half of the year, while the Four Little Dragons combined had 5.4 billion yuan in revenue and a loss of 400 million yuan. Cambricon's profit in half a year is more than the Four Little Dragons' total revenue!
- Reason for the gap: It's not about technology (Muxi even has a higher gross margin than Cambricon); it's about scale. Cambricon's larger revenue spreads its R&D costs, while the Four Little Dragons' higher R&D expenses (up to 40% of revenue) lead to losses.
- Implication: Cambricon also lost money in 2024 but made a sudden turnaround in 2025. Investors in the Four Little Dragons' stocks are betting on them repeating Cambricon's path to profitability.
3. Stock Price Volatility: Breaking the Bubble and Pressure from Share Restrictions
The initial excitement after listing has faded, and now we're in a period of calmness, even pain.
- Stock price drops: Moore Threads' and Muxi's stock prices have dropped significantly, showing that the market is no longer just interested in the concept but in actual performance.
- Share restrictions: Early investors (VC/PE) and original shareholders can now sell their shares, which could put pressure on prices.
- Moore Threads: Share restrictions will end on September 7th, with institutional investors selling aggressively, causing a price drop. There's an even larger sale in December (39% of the total shares).
- Muxi: Share restrictions will end on September 17th, and its smaller market size makes it more vulnerable to price fluctuations.
- Impact: Early investors bought shares at low prices and now have substantial profits, creating a strong incentive to sell, which will continue to pressure prices.
- Financing: Both companies plan to issue H shares in Hong Kong, which will increase share supply and dilute existing shareholders' equity.
4. The Technical Debate: CUDA Compatibility vs. Custom Architecture
This is a critical decision that determines their success:
- Mainstream approach (Moore Threads, Muxi, Beren): Compatible with CUDA, allowing easy integration with existing software.
- Advantages: Lower migration costs and easier access to customers.
- Disadvantages: Long-term dependence on NVIDIA's ecosystem, with potential patent and integration risks.
- Unique approach (Suoyuan): Custom architecture (DSA), not compatible with CUDA.
- Advantages: Potential for higher performance, but requires customers to rewrite their software, increasing costs.
- Current situation: Suoyuan's success depends on Tencent's support for adaptation. Without it, it's difficult for other customers to adopt this approach.
5. The Ultimate Question: Are These Companies Worth Their Prices?
- Profitability: All four companies are still losing money despite revenue growth.
- The high market valuation assumes they can achieve profitability like Cambricon through scale.
Key indicators: Whether they can maintain a gross margin of 40%-50% or more, and whether they can turn positive cash flows.
- Customer dependence: Can they reduce reliance on a single giant? Cambricon is diversifying, while Suoyuan is concentrating, posing different risks.
- Conclusion: In the short term, stock prices are volatile due to share restrictions and market sentiment. In the long term, only companies that can create a sustainable revenue, gross margin, and cash flow will survive.
- Warning: High sales don't necessarily equate to profitability. If gross margins don't improve or customer bases worsen, high revenue won't be meaningful.
**Advice for Investors:**
1. Don't just look at market value: Is a 600-billion-yuan market value a bubble or a reflection of real value? Look at net profit after deducting non-recurring expenses. All companies are still losing money, indicating they're in a growth phase.
2. Be cautious of share restrictions: Stock prices of Moore Threads and Muxi may be pressured by share sales by early investors in the coming months.
3. Watch Cambricon's impact: If Cambricon continues to profit, it will influence the overall sector's valuation; if its performance falls short, it could shake the valuation of the Four Little Dragons.
4. Long-term perspective: Domestic GPUs are a national priority, but the winner will depend on future financial reports, especially changes in gross margins and cash flows.
In summary: The game has begun, and while the stakes are high, the key to success lies in sustainable revenue, gross margins, and a diversified customer base.