Moore Threads Technology Co., Ltd. (688795) has become the undisputed "King of IPOs" in China's A-share market this year. On December 5, its stock price skyrocketed to 650 yuan per share at opening, marking a staggering 468.78% surge. The company's market capitalization briefly exceeded 300 billion yuan, with early investors potentially earning up to 270,000 yuan per lot if sold at peak. By market close, shares settled at 600.50 yuan, still up 425.5% from its IPO price of 114.28 yuan, with trading volume hitting 15.3 billion yuan—the highest single-day turnover across Chinese markets.
The company's roots trace back to NVIDIA (NVDA). Founded in June 2020 by Zhang Jianzhong, a 14-year veteran who led NVIDIA China to dominate 80% of the country's discrete GPU market, Moore Threads assembled an "NVIDIA Dream Team" with core members from the American chip giant. Unlike AI-focused startups, Zhang pursued a "full-featured GPU" strategy capable of handling graphics rendering, AI computing, and scientific calculations through its proprietary MUSA unified architecture.
Moore Threads' growth trajectory has been meteoric: securing billions in funding within a year of operation (2021), launching its first full-featured GPU, and completing its MUSA architecture by 2022. By 2024, annual revenue reached 438 million yuan, earning it "Beijing Unicorn" status. The 2025 IPO—submitted June 30 and listed December 5—set a capital markets record for speed.
However, challenges remain. Technologically, Moore Threads' 12nm chips lag behind NVIDIA's 4nm process by 2-3 generations, with AI computing efficiency at just 1/3 to 1/10 of NVIDIA's. Its 100,000-strong developer community pales against CUDA's 4 million users. Financially, cumulative losses exceeded 5 billion yuan from 2022-2024, with R&D spending (3.8 billion yuan) dwarfing total revenue. The company projects profitability no earlier than 2027.
Market-wise, while NVIDIA dominates 90%+ of AI training with global clients, 56% of Moore Threads' revenue comes from a single customer, primarily in government sectors. Its consumer GPU market share is merely 2%. Even compared to domestic peer Cambricon, Moore Threads' projected 2025 revenue (1.5 billion yuan) trails Cambricon's 7 billion yuan target.
Yet Moore Threads' long-term potential lies in its "full-stack computing platform" strategy. The MUSA ecosystem's CUDA compatibility lowers developer migration costs, while its versatility suits emerging needs like industrial digital twins (requiring AI+rendering+physics) and autonomous driving (perception+modeling+decision-making). State-backed import substitution programs—mandating 70% domestic GPU adoption in critical sectors by 2027—could unlock billion-yuan contracts.
Key milestones include achieving mass production of next-gen GPUs by 2027, matching NVIDIA in select performance metrics by 2030, and ultimately establishing China's autonomous GPU ecosystem. For now, Moore Threads' valuation reflects the market's bet on China's computing self-sufficiency—a high-stakes wager where technological ambition meets financial reality.