Moore Threads Technology, often dubbed China’s “first domestic GPU stock,” saw its share price surge further on Wednesday, breaching the $114 mark for the first time (870 yuan) and setting a new record high since its market debut last Friday. At one point during intraday trading, the stock rose more than 19%.
As of Wednesday morning, Moore Threads shares peaked above $124 (870 yuan), representing a gain of over 600% from its IPO price. The company’s intraday market capitalization surpassed $57 billion (400 billion yuan). Investors who were allocated shares in Moore Threads’ IPO and held them until today’s session could realize paper gains of more than $38,000 per lot at the intraday high.
The Beijing-based artificial intelligence chip designer is set to unveil its latest GPU architecture next week. Founder, chairman, and CEO Zhang Jianzhong will present the company’s fifth-generation GPU at the inaugural MUSA Developer Conference on December 19 and 20 in Beijing, according to a post on Moore Threads’ official WeChat account.
MUSA, or Meta-computing Unified System Architecture, is the firm’s proprietary platform, designed as a Chinese-developed alternative to Nvidia’s CUDA platform, which enables developers to harness GPU computing power for AI and other high-performance applications. During his keynote, Zhang will outline the MUSA roadmap, highlighting full-stack development capabilities for enterprise users.
The conference reflects Moore Threads’ strategy to expand both its hardware and software capabilities while growing its developer community amid rising competition from domestic GPU makers and Nvidia’s renewed push into the Chinese market with its H200 chips following U.S. approval.
Moore Threads’ confidence follows a successful IPO in Shanghai, which raised approximately $1.1 billion (8 billion yuan), ranking as the mainland’s second-largest public offering this year. Its stock soared 468% on debut, closing Wednesday at $105 (735 yuan), demonstrating strong investor appetite for a firm seen as China’s leading competitor to Nvidia.
The company’s previous GPU architectures include Sudi, Chunxiao, Quyuan, and Pinghu, the latter introduced last year to support training and inference of AI models with trillions of parameters. Pinghu-based GPUs have been employed in pre-training advanced large language models, including those from China’s DeepSeek, the firm previously said. Moore Threads is also a founding member of the Model-Chips Ecosystem Innovation Alliance, established in July to promote the adoption of locally developed AI processors.
Founded in 2020, Moore Threads has pledged to release a new-generation GPU annually to support China’s push for technological self-sufficiency and AI development.
“Our goal is to become a leading GPU player with international competitiveness,” Zhang, formerly Nvidia’s China chief, told the Shanghai Securities Journal last week.
