Summary of Key Points
MooreThread is a domestic GPU company that aspires to become the "Chinese version of NVIDIA." It plans to list on the Hong Kong Stock Exchange (HKEX) less than a year after its initial public offering (IPO) on the A-share market, becoming the first domestic GPU company with a dual listing on both A-share and HKEX markets. The company reported revenue of 1.736 billion yuan in the first half of the year, which is a significant increase compared to last year's total revenue. However, there are several underlying issues: it has negative cash flow of 2.169 billion yuan (meaning it spends more money than it earns), a loss of 150 million yuan after deducting non-recurring items (it only avoids losses thanks to government subsidies), a highly concentrated customer base (the top five customers account for 98% of its revenue), virtually zero overseas revenue, and has spent nearly 6 billion yuan on research and development (R&D) despite having less than 1% market share.
The main reason for its haste to list on the HKEX is a lack of funds. It also wants to take advantage of the current boom in the AI chip industry to raise more capital and provide shareholders with an opportunity to realize their investment gains. For China's chip industry, MooreThread's commitment to developing "full-featured GPUs" has strategic value. However, whether it can survive depends on its ability to convert its technology into actual revenue and market share.
1. Rapid Revenue Growth, but Negative Cash Flow and High Inventory Levels
MooreThread's revenue increased by 147% year-over-year in the first half of the year, which seems impressive. However, a closer look at its financial statements reveals problems: its cash flow from operating activities was -2.169 billion yuan, a 86% decline from the previous year. Its inventory amounted to 3.55 billion yuan, accounting for one-fifth of its total assets. This is due to its reliance on large customers for bulk purchases and intelligent computing projects, which result in slow payment receipts. Additionally, it needs to stockpile raw materials in advance, leading to a large amount of cash being tied up. With such high inventory levels, if there are rapid changes in chip technology or a drop in raw material prices, the inventory could become worthless, significantly impacting the company's performance.
Moreover, its reported profits are misleading: its net profit attributable to shareholders in the first half of the year was a loss of 11.56 million yuan, although this figure includes non-recurring gains of 88 million yuan from government subsidies and 59 million yuan from investments. Excluding these one-time earnings, the actual loss was 150 million yuan. In other words, the company is still relying on subsidies and investments to avoid losses.
2. Highly Concentrated Customer Base, Leading to Volatile Performance
MooreThread's growth is heavily dependent on a few key customers. In the first half of 2026, the top five customers accounted for 98.29% of its revenue, with one customer, R, accounting for 56.63%. This means that even a slight reduction in sales from this customer could cause a significant drop in the company's performance. For example, the company's revenue in the third quarter of 2025 was only 82.83 million yuan but soared to 700 million yuan in the fourth quarter. This fluctuation is risky for a listed company, as investors may feel that its success is at the mercy of these few customers and be reluctant to invest for the long term.
3. Calling for Internationalization, but with Zero Overseas Revenue
MooreThread claims it aims to "deepen its internationalization" by listing on the HKEX, but its overseas revenue is negligible. In 2025, it only generated 1.9 million yuan from overseas sales (0.13% of total revenue), and this figure dropped to zero in the first half of 2026. This is because the company was added to the US "Entity List" in 2023, preventing it from accessing US-made raw materials and technology tools, as well as entering the largest GPU markets in Europe and America. Although there have been rumors of potential cooperation with Middle Eastern investors, nothing has materialized yet. In contrast, NVIDIA generates approximately 60% of its revenue from overseas markets.
4. Nearly 6 Billion Yuan in R&D Expenses with Less than 1% Market Share
MooreThread has invested nearly 6 billion yuan in R&D but holds less than 1% of the domestic GPU market share and even fails to make it into the top ten in terms of AI accelerator shipments. The GPU industry is not only about hardware but also about the software ecosystem. NVIDIA has built a strong CUDA ecosystem over the past 20 years, making its tools widely used by developers. MooreThread's own MUSA architecture needs significant time and investment to gain traction among AI frameworks and game engines. Domestic competitors, such as Huawei's Ascend series and companies like Cambricon and WallRun Technology, are also competing for market share. MooreThread's attempt to develop full-featured GPUs could lead to a dispersion of resources, preventing it from gaining a competitive advantage in either area.
5. Hurrying to List on the HKEX: What’s the Real Motive?
The most immediate reason for MooreThread's move to the HKEX is a financial need; its ongoing negative cash flow and high R&D expenses mean that the funds raised from the A-share IPO may not last long. However, there is also a strategic aspect: the current AI chip boom makes it easier to attract capital from the market. If the trend fades, financing will become more difficult. Additionally, listing on the HKEX could attract international investors, such as Middle Eastern sovereign funds, and provide shareholders with an exit option (Pre-IPO investors have been waiting for six years to realize their investment gains). Although its overseas revenue is zero, listing on the HKEX can add credibility to its internationalization efforts.
For China's chip industry, MooreThread's efforts are valuable as it fills a gap in the market for full-featured GPUs and puts pressure on NVIDIA. However, whether it succeeds depends on its ability to convert its technology into sustainable revenue and market share. After all, the funds raised through listing must be effectively utilized to develop products that can attract customers.
Conclusion
MooreThread's story serves as a reflection of the challenges faced by domestic chip companies seeking to compete with giants. While their efforts are admirable, their survival depends on practical capabilities—specifically, their ability to generate revenue through market-ready products, rather than relying on subsidies or impressive rhetoric. In the highly competitive chip industry, having a vision alone is not enough; success requires the ability to turn that vision into real profits.