'Star Market Daily' reported on November 9th (by reporter Li Mingming) that the GPU unicorn Moore Thread has new developments.

'Star Market Daily' reporters found through a search of the National Enterprise Credit Information Publicity System that Moore Thread Intelligent Technology (Beijing) Co., Ltd. (referred to as 'Moore Thread') changed its market entity type from other limited liability company to another joint-stock company (non-listed) on October 28th. The company's registered capital increased from 24.413 million yuan to 0.33 billion yuan.

A knowledgeable source told 'Star Market Daily' reporters that Moore Thread has completed the shareholding system reform, is preparing for listing, and will soon start listing guidance. The STAR Market may be its primary target.

Established in October 2020, Moore Thread, led by legal representative Zhang Jianzhong, is an innovative enterprise with independent intellectual property rights in full-function GPU. Over the past four years, the company has iterated through four generations of chips. Last year, it set up the first domestic full-function GPU thousand-card intelligent computing cluster. The ten thousand-card intelligent computing cluster is currently under development.

According to data from the National Intellectual Property Administration, as of October 2024, Moore Thread has obtained 425 authorized patents, ranking among the top GPU companies in terms of the number of authorized patents in China.

The core team of Moore Thread is from NVIDIA, with founder Zhang Jianzhong serving as the former General Manager of NVIDIA China and Vice President of NVIDIA globally.

According to Cailianchuang Venture Investment Data, Moore Thread has gone through six rounds of financing since its establishment. It has attracted many well-known investors, including Sequoia China, Shunwei Capital, Tencent Investment, Zhongguancun Science City, etc. When it completed a 1.5 billion yuan Series B financing in December 2022, its valuation had already reached 24 billion yuan, making it one of the unicorn companies.

The current demand for computing power continues to increase, and for GPU companies, while facing challenges, it also brings new opportunities.

In 2023, Moer Thread launched the integrated Kuage KUAE smart computing cluster, which can be extended to thousands and tens of thousands of cards to build advanced computing infrastructure for large-scale models and general artificial intelligence. The cluster has completed training and inference testing of large models with different parameter magnitudes based on the Kuage smart computing cluster, collaborating with many domestic large model enterprises such as Zhipu AI, Zhiyuan Research Institute, Wumen Xinqiong, Dipu Technology, Rui Le Smart, and Real Smart.

At this year's Financial Street Forum annual meeting, the company signed on-site contracts with Shanghai Pudong Development Bank, Postal Savings Bank of China, and Bank of Beijing for the research project of the domestically produced full-featured GPU Wan Card universal smart computing cluster by Moer Thread.

A person in charge of Moer Thread told the Science and Technology Innovation Board Daily reporter that the company will continue to optimize GPU products to support the training and inference of large-scale models.

Recently, domestic GPU unicorns have witnessed an IPO boom.

In August, Suiyuan Technology signed a listing counseling agreement with China International Capital Corporation, officially starting the A-share IPO process; in September, AI chip unicorn Binren Technology filed for listing counseling registration with the Shanghai Securities Regulatory Bureau.

An industry insider told the Science and Technology Innovation Board Daily reporter that the domestic GPU chip market will experience strong growth in the coming years, with domestic GPU companies also moving towards technological innovation and product diversification. However, the research and development investment in AI chips is substantial, and it is difficult to sustain a single investment financing model. Therefore, domestic GPU unicorns are considering going public to get more financial support. But companies still need to enhance their technological innovation capabilities, explore new application scenarios, strengthen market adaptability, and expand their business models.