Li Jun of Sugon Discusses China's Computing Industry at Wuzhen: From Single-Point Catch-Up to Systemic Restructuring
Li Jun, Chairman of Sugon, proposed at the 2026 World Internet Conference Wuzhen Summit that China's computing industry needs to shift from single-point technology catch-up and product substitution to systemic restructuring, using system-level innovation to offset single-point shortcomings such as advanced processes and high-bandwidth memory. He introduced that the six core chips of the fully domestic 100,000-card AI supercluster Sugon 8000 have reached the international advanced level overall, with some products reaching the internationally leading level; the National Supercomputing Internet platform has nearly 2 million registered users, and the Guanghe Organization has gathered more than 6,000 partners and completed over 15,000 adaptation and optimization projects.
Body · AI translation
Li Jun Talks About China's Computing Industry in Wuzhen: Moving from Single-Point Catch-Up to Systematic Restructuring Source: Semiconductor Horizons Published: 2026-11-04 16:21 Computing Power Figures Generate Poster The competition in future computing power is not merely a competition over the peak performance of a single chip, but rather a contest of system capabilities to efficiently organize massive resources such as computing, storage, networking, and power.
On November 4, Li Jun, Chairman of Sugon, was invited to attend the 2026 World Internet Conference Wuzhen Summit and delivered a keynote speech at the Internet Entrepreneurs Forum. He pointed out that in the face of a new round of computing power demand brought by artificial intelligence, China's computing industry needs to move from single-point technology catch-up and product substitution toward systematic restructuring, make up for shortcomings through system-level innovation, and unite industry forces through open collaboration to form computing power competitiveness for the AI era.
Breaking Through in Computing Power Cannot Focus Only on a Single Chip
As large models continue to evolve, intelligent agents drive growth in inference demand, and AI4S continuously expands scientific research applications, global demand for computing power continues to rise. At the same time, problems such as the shortage of high-end computing power, a weak basic software ecosystem, and low computing power utilization efficiency coexist, becoming real challenges for the development of domestic computing power.
Li Jun believes that China still faces constraints in key links such as advanced processes and high-bandwidth memory, but it also has rich application scenarios, a relatively complete industrial chain, strong engineering capabilities, and advantages such as energy supply and the open-source ecosystem. To leverage these advantages, it is not enough to rely only on catching up in the performance of a single chip; it is also necessary to focus on overall system efficiency and explore a development path of systematic innovation.
"The competition in future computing power is not merely a competition over the peak performance of a single chip, but rather a contest of system capabilities to efficiently organize massive resources such as computing, storage, networking, and power."
Li Jun proposed using system-level innovation to offset single-point shortcomings and using open collaboration to pool industry forces. On the one hand, promote coordinated optimization of links such as computing, storage, networking, power supply, and cooling to improve overall system efficiency; on the other hand, promote collaborative innovation among chips, basic software, models, and industry applications to transform industrial chain advantages into actual computing power supply capacity.
Chips, Systems, and Services Work Together
Regarding the specific path for systematic restructuring of the industry, Li Jun proposed to "focus on computing chips, computing systems, and computing services, and work with ecosystem partners to jointly develop China's computing industry." From core technology to system architecture and then to computing services, the three support each other and drive the transformation of technological innovation into application value.
On the chip side, Sugon and related enterprises continue to innovate in key areas such as CPUs, intelligent computing chips, and high-speed interconnects. Li Jun introduced that the six core chips applied in the fully domestic 100,000-card AI supercluster Sugon 8000 have reached the international advanced level overall, and some products have reached the international leading level. At the same time, they are actively exploring technologies such as advanced packaging and silicon photonic devices to further enhance computing and interconnect capabilities under existing process conditions.
On the system side, Sugon is advancing integrated innovation in chips, computing, storage, networking, power, and cooling under the direction of "super nodes + super clusters." Citing Sugon 8000 as an example, Li Jun pointed out that through large-scale system collaboration, high-speed interconnection, storage optimization, and open compatibility, the actual application capability of domestic computing power has made notable progress. He also emphasized that whether domestic chips are good to use cannot be judged solely by single-card metrics or past impressions, but should be based on performance in real applications, and he welcomed the industry to experience it firsthand through Sugon 8000 and the National Supercomputing Internet.
On the service side, Li Jun emphasized the need to develop the computing service industry and promote the transformation of computing power resources into application value. As of now, the National Supercomputing Internet platform has nearly 2 million registered users and is connecting computing power resources, software, and scientific research needs through platform-based services. Around AI4S, Sugon is exploring a service model of "Supercomputing Internet + Super Scientific Computing Intelligent Agent" to make domestic computing power easier to use and more inclusive.
Only with Open Division of Labor Can Industrial Synergy Be Formed
System innovation cannot be separated from the collaboration of the industrial ecosystem. At present, the domestic AI software stack still has shortcomings in areas such as toolchains, compatibility and adaptation, and the application ecosystem, and complex problems in the industrial chain are also difficult for a single enterprise to solve independently. Li Jun called on enterprises to leverage their respective strengths, reduce duplicate investment and low-level competition through professional division of labor and open cooperation.
At present, the Photon Organization has gathered more than 6,000 partners and carried out more than 15,000 adaptation and optimization tasks. Through software and hardware adaptation, cross-layer technical collaboration, and application innovation, it promotes open compatibility among different technical architectures and enables the technical advantages of upstream and downstream players in the industrial chain to form synergy.
"Closing the door and doing everything by oneself absolutely will not work." Li Jun emphasized that independent innovation does not mean closed development. Facing the intelligent era, China's computing industry must adhere to independence, openness, security, and inclusiveness, rely on open platforms such as the World Internet Conference, strengthen international industry-university-research cooperation, and jointly build and share an open ecosystem with the world.
From substitution to restructuring, what domestic computing power needs to break through is not only single-point technical bottlenecks, but also the formation of industrial capabilities for the coordinated development of chips, systems, services, and ecosystems. Improving overall efficiency through system innovation, unleashing application value through computing services, and cultivating the industrial ecosystem through open cooperation are precisely important paths for China's computing industry to enhance long-term competitiveness. Sugon will also continue to work with industry partners to promote the systematic restructuring of the computing industry through open collaboration and provide a more solid computing power foundation and more easy-to-use computing services for the AI era.
This article is reproduced from media reports or online platforms and represents the author's personal stance or views. Our reproduction is solely for sharing and does not mean that we endorse or agree with it. If the source attribution is incorrect or your lawful rights and interests are infringed, please contact customer service promptly, and as a neutral platform service provider we will promptly correct, delete, or handle it in accordance with the law. Comments No user comments yet Latest News More Related Recommendations
$517.7 billion in a single quarter! Global semiconductor sales achieve three consecutive increases 2026-11-04 16:39
The world's second-largest photoresist company is about to be sold 2026-11-04 16:26
Saiteng Co., Ltd.: Semiconductor inspection equipment has been delivered in batches to overseas storage customers 2026-11-04 16:11