Moore Threads

3wks agoupdate 143 0 0

Specialized in the development of full-featured GPU chips, the company provides high-performance computing acceleration solutions for a wide range of fields, including artificial intelligence, graphics rendering, and scientific computing.

Location:
China
Language:
zh,en
Collection time:
2025-08-30
摩尔线程Moore Threads

About Moore Threads

Founded in June 2020 and headquartered in Beijing, Moore Threads Intelligent Technology (Beijing) Co., Ltd. is a company that specializes in the research and development of full-featuredGPU chipThe company is a high-tech enterprise of integrated circuits and its related products. With its self-developed full-featured GPUs as the core, the company is committed to providing computing acceleration platforms for high-performance computing fields such as artificial intelligence (AI), digital twins, scientific computing, etc., and is mainly engaged in the research, development, design and sales of GPUs and related products. In the context of increasingly fierce global competition in science and technology, GPUs, as key computing chips, are widely used in a variety of fields, and have become one of the core driving forces to promote the development of the digital economy. With its profound technical accumulation and forward-looking strategic layout, Moore Threads has risen rapidly in the field of domestic GPUs and become a leading company in the industry.

Milestones

Start-ups and breakthroughs (2020-2021)

In June 2020, Moore Threads was formally established, bringing together a group of core talents from international semiconductor giants such as NVIDIA and AMD to form a professional team with rich experience in GPU R&D. These talents have an average of more than 15 years of experience in the GPU industry, laying a solid foundation for the company's technological innovation and product development. These talents have an average of more than 15 years of experience in the GPU industry, laying a solid foundation for the company's technological innovation and product development. in November 2021, Moore Threads announced the successful development of the first full-featured GPU, marking the company's major breakthrough in the development of GPU technology, and filling the gaps in the field of full-featured GPU in China.

Product Ecology Building (2022-2023)

In March 2022, Moore Threads debuted the new MUSA Unified System Architecture, which integrates four major engines for AI compute acceleration, graphics rendering, physics simulation and scientific computing, and ultra-high-definition video coding and decoding, and is a core achievement of the company's technological innovation. At the same time, the company also launched the MTT S50, a desktop-class graphics card for PCs and workstations, the MTT S2000, built for data centers, and the GPU physics engines AlphaCore and DigitalMEdigital personIn November, Moore Threads launched its second full-featured GPU chip "Spring Dawn", and based on it, the company built the industry's first domestic game graphics card MTT S80 and server-oriented MTT S3000, as well as the meta-computing all-in-one machine MCCX. In addition, the Company released a series of GPU software stacks and application tools to further improve the product ecosystem and enhance the market competitiveness of its products.

Technology Iteration and Ecological Expansion (2023-2024)

In 2023, Moore Threads continued to maintain its rapid development. in March, the company was recognized as a "Specialized, Specialized and New" enterprise in Beijing. in May, the company released a series of new products and technology updates, covering important GPU application scenarios, such as digital office, entertainment and creativity, AI and cloud computing, as well as meta-universe, etc. in September, the company released a new generation of full-featured GPU chips, "QuYuan". In September, the company released a new generation of full-featured GPU chip "QuYuan", and based on "QuYuan", launched the AI training and pushing integrated computing card MTT S4000 and Moore Threads KUAE (KUAE) 1,000-card smart computing cluster, which is mainly used for the training of large models with hundreds of billions of parameter scales as well as reasoning. In addition, Moore Threads also launched MCCPlatform, an arithmetic management platform designed for arithmetic clusters, MCCFlow, an application development platform, and MusaChat, Moore Threads' self-developed large-language model, and in December, Moore Threads' KUAE Intelligent Computing Center, the first nationally-produced GPU KUAE model training platform, was inaugurated in Beijing. The unveiling ceremony was successfully held in Beijing, announcing that the first large-scale arithmetic cluster based on domestic full-featured GPUs had officially landed in China.

Listing Sprint and Ecological Deepening (2025 to present)

In 2025, Moore Threads ushered in a new stage of development. on June 18, the company's IPO counseling status was officially changed to "counseling acceptance", marking that the listing process has entered the substantive review stage. on June 30, the company's application for IPO on Science and Technology Creation Board (SCCB) was accepted. on July 8-9, Moore Threads participated in the 2025 From July 8 to 9, Moore Threads participated in the 2025 China Mobile Industry Investment and Eco-Cooperation Conference, showcasing its general computing acceleration platform based on domestic full-featured GPUs and its innovative achievements, includingAI macromodelMoore Threads has joined China Mobile's "Fertile Ground Program" and will cooperate more deeply with China Mobile and its ecological partners in technology research and development and application landing. At the same time, Moore Threads has joined China Mobile's "Fertile Ground Program", and will launch more in-depth cooperation with China Mobile and its ecological partners in technology research and development and application landing, and jointly promote the development of the artificial intelligence industry.

Products & Services

Consumer Graphics Cards

Moore Threads has made significant breakthroughs in the field of consumer graphics cards. 2022 saw the launch of the MTT S80, the first consumer graphics card in China to support Windows operating systems and DirectX 11/12 graphics libraries, with performance specifications comparable to those of the NVIDIA RTX 3060. After more than two years of continuous technology iteration and up to 32 driver updates, MTT S80 has now been upgraded to be the first and only domestic graphics card to support DirectX12 in China, capable of smoothly running the latest 3A-grade game masterpieces such as Black Myth Wukong and other games, which are highly favored by gamers. In addition, the company has also launched MTT S70 and other gaming graphics cards to meet the needs of different user groups.

Data Center Graphics

In the data center field, Moore Threads has launched a number of graphics cards designed for server applications. MTT S2000 is a graphics card specifically built for data centers with powerful computing power and efficient energy efficiency ratio, which is able to meet the demand for high-performance computing in data centers. MTT S3000 further enhances the performance by supporting a single card with a maximum of 32 vGPU slices, and concurrently supporting 32 VMs MTT S3000 further improves performance by supporting a maximum of 32 vGPU slices on a single card and concurrently supporting 32 VMs. MTT S5000 and MTT S4000 are high-performance graphics cards for the AI smart computing market, which can provide powerful arithmetic support for large model training and inference.

Intelligent Computing Cluster Solutions

Moore Threads has launched the "KUAE" full-stack solution for the Intelligent Computing Center, which supports 10,000-calorie scale-up capability. The solution integrates advanced GPU chip, high-speed interconnection technology and intelligent management system, which can provide stable and efficient computing power support for DeepSeek and other large models with hundreds of billions of parameters to be pre-trained. The efficiency of the 1,000-card smart computing cluster built based on MTT S5000, its latest generation AI flagship product, exceeds the computational efficiency of foreign GPU training clusters of the same generation of the same scale, demonstrating strong technical strength and market competitiveness.

Software & Services

Moore Threads also provides a series of software and services, including the MUSA Developer Suite, cloud-native sGPU technology, and the meta-universe platform MTVERSE. These software and services provide developers with a convenient development environment and rich application scenarios, promoting the prosperous development of the GPU ecosystem. Meanwhile, the company also provides comprehensive technical support and after-sales service to ensure that customers can make full use of Moore Threads' products and services to maximize business value.

core technology

MUSA Unified System Architecture

MUSA architecture is the core achievement of Moore Threads' technological innovation, and it is the first full-featured GPU architecture in China that realizes the simultaneous support of AI intelligent computing, graphics acceleration, and physical simulation on a single chip. The architecture integrates four engines of AI acceleration, graphics rendering, physical simulation and scientific computing, and ultra-high-definition video coding and decoding, which provides strong technical support for Moore Threads' products. Based on the MUSA architecture, Moore Threads' GPU products are highly versatile and flexible, and are able to simultaneously meet the needs of multiple application scenarios such as graphics rendering, AI computing, and scientific computing.

Full-featured GPU chip design

Moore Threads has deep technical accumulation in full-featured GPU chip design. The company adopts an innovative multi-engine, scalable GPU architecture and builds a globally shared pool of computing, memory and communication resources through hardware resource pooling and dynamic resource scheduling technology. This design not only breaks through the limitations of traditional GPUs with a single function, but also significantly improves resource utilization while guaranteeing versatility. Its parameterized configuration and scalable architecture allows for rapid tailoring of optimized chip configurations for target markets, significantly reducing the development cost of new chips.

High-speed interconnection technology

In Smart Computing clusters, high-speed interconnect technology is the key to ensure efficient collaboration of large-scale GPUs. Moore Threads has created an original ACE asynchronous communication engine, which reduces the loss of computing resources by 15%. At the same time, the company has also developed MTLink2.0 interconnection technology, which provides a bandwidth of 60% higher than the domestic industry average, laying a solid foundation for large-scale cluster deployment. These high-speed interconnection technologies enable Moore Threads' Smart Computing cluster to realize efficient collaboration of thousands of nodes, significantly improving the efficiency of large model training.

Software ecological compatibility

Moore Threads focuses on the compatibility construction of the software ecosystem. The company's self-developed MUSA architecture has good compatibility with CUDA and other mainstream programming models, making it convenient for developers to migrate existing CUDA applications to the MUSA platform. At the same time, Moore Threads also provides a wealth of development tools and resource libraries, including MUSA compiler, AI arithmetic libraries, general-purpose computing libraries, and so on, to help all kinds of applications to quickly land. This compatibility building not only reduces the migration cost of developers, but also promotes the prosperous development of Moore Threads ecosystem.

development prospect

Market demand continues to grow

With the continuous evolution of cutting-edge technologies such as AI big models, digital twins, autonomous driving, and virtual reality, GPUs are becoming the core engine of the next-generation arithmetic infrastructure. According to Frost & Sullivan's forecast, the global GPU market size is expected to reach 3,611.974 billion yuan in 2029, of which, China's GPU market size will reach 1,363.578 billion yuan, and the global share is expected to increase from 30.8% in 2024 to 37.8% in 2029. Moore Threads, as a leading domestic full-featured GPU enterprise will fully benefit from the continuous growth of market demand and usher in a broad space for development.

Technological innovation to promote product upgrading

Moore Threads will continue to maintain its high-speed technology iteration capability, launching a new GPU chip every year and continuously improving the performance and application scenarios of its products. The company will continue to deepen the research and development of MUSA architecture, optimize chip design, high-speed interconnection technology and software ecology, and promote the full-featured GPU to make new breakthroughs in terms of versatility, flexibility and energy-efficiency ratio. Meanwhile, Moore Threads will also actively explore new application scenarios and technology areas, such as quantum computing and photonic computing, to lay a solid foundation for future development.

Ecological construction for industrial synergy

Focusing on ecological construction, Moore Threads has built an open industrial ecosystem through the layout of the whole chain of "chip-software-service". Moore Threads has established good cooperative relationships with many head operators, large state-owned banks, and head model startups to jointly promote the development of the artificial intelligence industry. In the future, Moore Threads will continue to strengthen cooperation with ecological partners, jointly explore new application scenarios and business models, and promote the collaborative development of the industry. At the same time, the company will also actively participate in the development of international standards and industry exchange activities to enhance the international influence of China's GPU industry.

Listed financing to help long-term development

Moore Threads has started the A-share IPO process, and plans to raise 8 billion yuan for the research and development of new-generation AI training and push all-in-one chips, graphic chips, AI SoC chips, and supplementing working capital. Listing financing will provide strong financial support for the company's long-term development, which will help the company to increase R&D investment, expand production scale and enhance market competitiveness. At the same time, the listing will also enhance the company's brand awareness and market influence, attract more excellent talents and partners for the company, and promote the company to realize leapfrog development.

data statistics

Relevant Navigation

No comments

none
No comments...