Moore Threads Leads China’s Chip IPO Wave with Record US$1.1 Billion STAR Market Listing

Moore Threads Technology, a leading Beijing-based graphics processing unit developer often dubbed the "Chinese Nvidia," has formally initiated the subscription process for its initial public offering on Shanghai’s Science and Technology Innovation Board, known as the STAR Market. The listing marks a pivotal moment for China’s semiconductor industry, as Moore Threads becomes the first of the country’s "four GPU dragons" to go public, capitalizing on a surge in domestic demand for artificial intelligence computing power amid ongoing U.S. export controls.

The IPO sets multiple records for the STAR Market in 2025. From application acceptance to approval, the process took a mere 88 days, the fastest on the board this year. With an offering price of RMB 114.28 per share and anticipated gross proceeds of approximately RMB 8 billion, the deal stands as the year's largest and highest-priced offering on the venue. The issuance costs alone, excluding VAT, reached RMB 424 million (US$58.6 million), signaling the sheer scale of the transaction.

This listing underscores the strategic convergence of market demand and state policy. As U.S. restrictions on advanced chips limit access to Nvidia Corp.’s high-end hardware, Chinese tech companies face an acute supply gap for AI training and inference capabilities. Concurrently, Beijing has elevated technological self-reliance to a national strategy, optimizing listing regulations on the STAR Market to facilitate access to capital for high-investment, research-intensive chipmakers.

While the listing highlights the sector's growth potential, Moore Threads faces the classic volatility of the semiconductor industry. The company has transitioned its revenue base toward AI computing products, which now account for the vast majority of its income, yet it continues to grapple with substantial net losses due to aggressive spending on research and development.

The "Nvidia of China" Ambition

The comparisons between Moore Threads and Nvidia are rooted in the company’s leadership and technical genealogy. Founder and CEO Zhang Jianzhong spent 15 years at Nvidia, rising to Global Vice President and General Manager of Greater China, where he helped expand Nvidia’s local market share from under 50% to 80% by 2020. Zhang left to establish Moore Threads in June 2020.

This DNA extends throughout the management and technical teams. Co-founder Zhou Yuan formerly served as Nvidia’s Senior Director of Market Ecosystems, while deputy general managers Wang Dong and Song Xuejun held senior sales roles at the US giant. Key technical personnel, including CTO Zhang Yubo, previously worked as GPU architects at Nvidia.

Leveraging this background, Moore Threads has adopted a "full-function GPU" technical route, distinct from competitors focusing solely on AI acceleration. The company aims to replicate Nvidia’s "one chip, multiple capabilities" model, developing its proprietary MUSA architecture to unify 3D graphics rendering, scientific computing, and AI training within a single environment. This strategy positions the company to service diverse sectors, including digital twins and the metaverse, rather than competing solely in the commoditized AI accelerator market. The company is also building its own software ecosystem to reduce reliance on Nvidia’s dominant CUDA platform, offering compatibility tools to ease developer migration.

Financial Pivot to AI and Capital Intensity

The explosive demand for Large Language Models (LLMs) has fundamentally altered the company's financial profile. According to the prospectus, 2024 marked a watershed moment where AI computing products replaced professional graphics cards as the primary revenue driver. In 2024, AI products generated RMB 336 million, representing 77.63%, accounting for 94.85% of revenue.

Gross margins for these AI products remain robust, recording 73.12% in 2024 and 69.37% in the first half of 2025. However, these figures contrast sharply with the company’s massive capital burn rate. R&D expenses totaled RMB 2.517 billion (US$348 million) between 2022 and the first half of 2025.

Consequently, Moore Threads operates at a steep loss. After deducting non-recurring gains and losses, the net loss attributable to shareholders was RMB 1.633 billion (US 225.8million) in 2024 and RMB 317 million (US 43.8 million) in the first half of 2025. As of June 30, 2025, the company reported accumulated uncovered losses of RMB 1.604 billion.The majority of the IPO proceeds—approximately RMB 7 billion—are earmarked for further R&D in next-generation AI training chips and graphics processors.

Concentration Risks and Market Dynamics

Investment risks remain pronounced, particularly regarding client concentration. The prospectus reveals that the combined revenue from the top five customers exceeded 80% in recent periods, reaching 98.29% in the first half of 2025. While common in the early stages of enterprise-facing semiconductor sales, this dependency exposes the company to significant volatility.

Moore Threads leads a pack of startups known as the "Four Little Dragons" of domestic GPUs, which also includes Iluvatar CoreX, Enflame Technology, and Biren Technology. Moore Threads' successful IPO is expected to accelerate listing plans for its peers.

The broader industry faces cyclical challenges driven by technical innovation and capital expenditure fluctuations. The sector is currently in a shorter cycle of rapid iteration, with data center GPUs updating every 12 to 24 months to meet AI "arms race" demands. However, geopolitical friction has created a strategic window for Chinese firms. By offering domestic alternatives, Moore Threads and its peers are competing in a tiered market, with full-function GPUs addressing broad versatility, while other architectures focus on specific high-performance computing or inference niche markets.