1. Introduction: The madness of the capital market and the calm of the chip industry
On December 5, 2025, when Moore Threads (688795.sh) was listed on the Shanghai Stock Exchange's Sci-Tech Innovation Board, the entire A-share market was in an uproar. The opening price on the first day was 650 yuan/share, soaring 468.78% from the issue price, with an intraday high of 688 yuan/share, and finally closing at 600.5 yuan/share, a single-day increase of 425.26%. On the first day of listing, the market value exceeded 282.3 billion yuan, making it the fifth largest enterprise by market value on the Sci-Tech Innovation Board.
"I have been trading stocks for 30 years, but my returns are not as good as getting a single lottery from Moore." The sigh of stockholders in the stock forum vividly depicts the spectacle of this capital carnival. However, behind this madness lies a more thought-provoking reality: this company, hailed by the market as 'China's version of Nvidia,' had a net loss of 724 million yuan as of the end of September 2025, with cumulative losses exceeding 5 billion yuan.
The lightning-fast review speed of 88 days and the stark contrast of a 700 million loss have made Moore Threads the most dramatic focus in the capital market in 2025. Its rise is either a glimmer of hope for breaking through in the domestic GPU chip industry or yet another feast of capital speculation. Can it replicate NVIDIA's business myth, or will it fall into the valuation trap common among domestic tech companies? This is the core issue that this article aims to explore in depth.
2. 88 days approval: The threefold logic behind the lightning speed
Moore Threads took only 88 days from IPO acceptance to approval, setting the fastest record for review on the Science and Technology Innovation Board. This "lightning speed" is the result of the resonance of three logics: "scarce targets + policy endorsement + industrial demand".
(1) Scarcity: Filling the gap in the A-share market
As the only company in China that can functionally benchmark with NVIDIA, Moore Threads has an irreplaceable scarcity in the A-share market. Unlike most GPU companies that focus on a single function (such as graphics rendering or AI acceleration), Moore Threads integrates multiple capabilities such as graphics rendering, AI computation, video processing, and scientific computing based on its independently developed MUSA architecture on the same chip architecture. It can be simply understood as: full-function GPU = AI chip + GPGPU + graphics chip.
This technical path not only makes Moore Threads one of the few domestic companies that can benchmark NVIDIA in product form but also allows full-function GPUs to have a wide range of application spaces—satisfying the needs of traditional PC and workstation fields while also playing an important role in emerging fields such as AI, big data processing, cloud computing, and digital twins.
(2) Policy dividends: Direct beneficiaries of the Science and Technology Innovation Board reforms
The rapid approval of Moore Threads is inseparable from the support of the Science and Technology Innovation Board's "1+6" reform policy. In 2025, the reform of the Science and Technology Innovation Board will establish a "Science and Technology Growth Layer" that allows unprofitable hard technology companies to go public and optimizes the review process, providing key support for strategic emerging industries such as artificial intelligence and integrated circuits.
As a national "specialized, refined, and new" small giant enterprise and an industry unicorn, Moore Threads focuses on the research, design, and sales of full-function GPU products, being one of the few domestic GPU companies that balance graphics rendering and AI computation. Since its establishment in 2020, it has aimed at the high technical threshold and strategically significant full-function GPU route, achieving technological breakthroughs that support AI computing acceleration, graphics rendering, physical simulation, scientific computing, and ultra-high-definition codec on a single chip, and is also the first domestic company to launch a GPU that supports the DirectX12 graphics acceleration engine.
(3) The urgent demand for domestic substitution driven by industrial necessity
In the context of intensified Sino-U.S. technological competition and stricter export controls on high-end GPUs, the localization of computing power has risen to a national strategy, with urgent demand for domestically controllable GPUs in key areas such as finance, energy, and government affairs. Data shows that the current market share of domestic GPUs in the domestic market is less than 1%, with huge room for substitution.
The listing of Moore Threads just provides capital markets with a core target to participate in this wave of domestic computing power substitution. Its products are approaching or have reached international advanced levels in some performance indicators, such as the MTTS80 graphics card's single precision floating point performance being close to NVIDIA's RTX 3060, providing domestic companies with a viable domestic alternative.
3. 700 million loss: An inevitable result under high investment and long cycles
Despite the rapid growth of Moore Threads' revenue being described as "rocket speed," rising from 46.08 million yuan in 2022 to 438 million yuan in 2024, with a compound annual growth rate of 208%, and reaching 702 million yuan in the first half of 2025, exceeding the total revenue of the previous three years, it is accompanied by a continuously widening loss scale.
(1) The intensity of R&D investment far exceeds the industry average
From 2022 to 2024, Moore Threads' R&D investment reached 1.116 billion yuan, 1.334 billion yuan, and 1.359 billion yuan respectively, with a cumulative investment of 3.81 billion yuan over three years, accounting for as much as 626.03% of the cumulative operating income during the same period. At the same time, due to the lag in revenue recognition for GPU products, customers need to evaluate the compatibility of product performance with business needs over the long term before making bulk purchases, resulting in the company's revenue growing from 46 million yuan in 2022 to 438 million yuan in 2024, but the absolute scale remains small, making it difficult to cover high costs. In addition, the company was listed on the "entity list" by the United States in 2023, and supply chain constraints further affected market expansion pace.
Compared to comparable companies in the same industry, Moore Threads has a relatively high amount of losses, primarily due to greater R&D investment intensity, focusing on the development of full-function GPUs, leading to higher business complexity and more significant cost pressures.
(2) High customer concentration and insufficient revenue stability
From 2022 to 2024, the sales of Moore Threads' top five customers accounted for 89.86%, 97.45%, and 98.16% respectively, and in the first half of 2025, it rose to 98.29%. The high customer concentration leads to significant revenue volatility risks. Although this is a common phenomenon in the AI chip industry, for Moore Threads, which is still in its early stages, changes in the purchasing rhythm of a single customer directly affect the revenue scale, making it difficult to cover fixed costs through diversified income, further exacerbating the loss situation.
(3) Long-term investment in industrial chain layout and ecological construction
In addition to the high investment in chip research and development, Moore Threads is also actively building the GPU industry ecosystem. Its independently developed MUSA unified system architecture covers a full-stack technology system including chip architecture, instruction set, programming model, software runtime library, and driver framework, requiring continuous investment of substantial resources for ecological construction and developer support.
As of June 2025, Moore Threads has obtained 514 authorized patents, including 453 domestic invention patents, establishing a solid intellectual property barrier. At the same time, the MUSA SDK developed by the company covers programming frameworks, accelerators, deployment tools, and other full development resources, with over 300 application adaptations completed, compatible with mainstream frameworks such as Python, PyTorch, and TensorFlow, and lowering the threshold for migrating the CUDA ecosystem through the MUSIFY tool.
4. The NVIDIA myth: A business miracle that is hard to replicate
To judge whether Moore Threads can replicate NVIDIA's business myth, we first need to understand the core logic of NVIDIA's success. NVIDIA's rise is essentially a perfect storm of "technological innovation + ecological construction + strategic positioning."
(1) The success code of NVIDIA
The long-termism of technological iteration: Since its establishment in 1993, NVIDIA has always adhered to the GPU technology route, accumulating for over 20 years, gradually expanding from graphics rendering to AI computing and high-performance computing, forming a deep technological barrier.
The construction of ecological moats: NVIDIA has built the largest GPU developer ecosystem in the world through the CUDA platform, with over 4 million developers and countless compatible applications. This ecological barrier makes it difficult for competitors to surpass in the short term.
The precision of strategic positioning: NVIDIA made early arrangements before the AI explosion by acquiring companies like Mellanox to improve its industrial chain layout, successfully capturing the surge in computing power demand brought by AI large model training.
(2) The comprehensive gap between Moore Threads and NVIDIA
Despite significant progress in some aspects, there are still comprehensive gaps compared to NVIDIA.
Technical generational gap: In terms of process technology, Moore Threads' current main products use a 12nm process, while NVIDIA has mass-produced the 5nm process H100/H200 series, with the generational gap in process technology directly leading to significant differences in performance and power consumption ratio. In terms of computing performance, the computing power of Moore Threads' high-end products is only about half that of NVIDIA's flagship models, and this gap is further amplified in cluster computing scenarios.
Ecological barriers: NVIDIA's CUDA ecosystem has accumulated for nearly 20 years, with a community of 4 million developers and countless compatible applications, while Moore Threads' independently developed MUSA architecture claims compatibility with CUDA, but actual testing shows that the compatibility rate with the PyTorch framework is only 68%, with insufficient support for long-tail operators, and the developer community size is only 100,000, meaning the cost for companies to migrate models to the MUSA platform will increase by 30%.
Financial strength: NVIDIA's revenue in the 2024 fiscal year reached $60.9 billion, with a net profit of $29.7 billion and R&D investment as high as $10.2 billion. In contrast, Moore Threads' revenue in the first three quarters of 2025 was only 785 million yuan, with a net loss of 724 million yuan. Although the intensity of R&D investment is high, the absolute scale remains vastly different from NVIDIA.
Market share: NVIDIA's global GPU market share exceeds 80%, holding an absolute dominant position in the AI training market. In contrast, Moore Threads' domestic market share is less than 1%, with most products still in small-scale validation stages.
5. Path to Breakthrough: The Real Choices of Moore Threads
Despite facing many challenges, Moore Threads is not without the possibility of breaking through. In the context of the wave of domestic substitution and the explosive demand for AI computing power, Moore Threads can achieve differentiated development through the following paths:
(1) Focus on domestic substitution and niche markets
In the short term, Moore Threads should avoid direct competition with NVIDIA in the high-end market, instead relying on policy support for domestic substitution dividends, primarily targeting areas such as government affairs and finance that have high requirements for security and controllability; in the consumer market, focusing on mid-range scenarios such as gaming and creative design to fill domestic gaps; in the AI computing field, avoiding the high-end training market and focusing on mid-range inference and edge computing scenarios where NVIDIA is limited.
(2) Strengthen ecological construction and reduce migration costs
Ecological construction is the key to the long-term development of Moore Threads. The company should continuously increase investment in MUSA ecological construction, improve the developer toolchain, enhance software compatibility, and reduce migration costs for enterprises. At the same time, by deeply binding with domestic operating systems and AI frameworks, it can form a domestically controllable technology stack to break free from dependence on CUDA.
(3) Deepen technological cooperation to break through supply chain bottlenecks
In the current international situation, Moore Threads should strengthen cooperation with domestic wafer manufacturing companies to jointly tackle advanced process technology challenges. At the same time, through innovative means such as Chiplet technology, it can break through advanced process limitations and improve performance and power consumption ratios.
(4) Optimize customer structure and reduce operational risks
Moore Threads should actively expand its customer base and reduce reliance on a single customer. By providing customized solutions to meet the personalized needs of customers in different industries, it can enhance customer loyalty and retention.
6. Conclusion: A rational view of "China's version of NVIDIA"
The listing of Moore Threads marks a new development stage for the domestic GPU industry. As the only company in China to achieve full-function GPU mass production and sales, Moore Threads has made significant progress in technological innovation, product layout, and ecological construction, making important contributions to the construction of the domestic computing power system.
However, we must also recognize that there is still a huge gap between Moore Threads and NVIDIA, and replicating NVIDIA's business myth is not an easy task. In the current capital market's frenzy for AI chips, it is even more necessary to maintain rationality and calmness, objectively assessing the true value and development potential of the enterprise.
For investors, attention should be paid to the core technological strength of enterprises, progress in ecological construction, and commercialization capabilities, rather than blindly following trends. For Moore Threads, the listing is just the starting point; the real challenge lies in continuously exerting efforts in technological iteration, ecological construction, and market expansion to achieve the leap from "domestic substitution" to "global leadership."
In the future, whether Moore Threads can become "China's version of NVIDIA" will depend not only on its own efforts but also on the overall development of the domestic GPU industry ecosystem. With the support of national policies and the drive of market demand, we have reason to believe that the domestic GPU industry is likely to achieve breakthroughs, bringing new choices to the global computing power market.
However, this is destined to be a long marathon that requires patience, perseverance, and continuous investment. For Moore Threads, a 700 million loss is not the end, but a necessary path to success. As NVIDIA founder Jensen Huang said: "Great companies are not born in prosperity, but grow in adversity." Whether Moore Threads can grow in adversity and become the true "China's version of NVIDIA" is something we will wait and see.