From Chain Expo to Frontrunners Summit Huang Renxun: Hope to provide more advanced chips to China, optimistic about China's advantages in AI applications and other fields.

On July 16th, after attending the third Chain Expo, NVIDIA CEO Huang Renxun went to Beijing front gate to participate in a media exchange meeting. During the over an hour long meeting, Huang Renxun discussed NVIDIA's technology strategy, industry competition, China market layout, and AI development, showing NVIDIA's optimism towards the future of AI in China and its emphasis on the Chinese market. The newly released RTX Pro GPU is suitable for scenarios such as digital twins, autonomous driving, and Siasun Robot & Automation. During this meeting, he clearly emphasized the importance of the Chinese market. Huang Renxun hopes to provide China with more advanced chips than the H20. Currently, H20 still has an advantage due to the inference capabilities and bandwidth of the Hopper architecture, but as technology continues to develop, products sold to China will continuously be upgraded. He also mentioned that the resumption of H20 sales in China depends on approval from the U.S. government, and restarting the supply chain will take time, but customer demand is strong. Huang Renxun also introduced the newly released RTX Pro GPU based on the Blackwell architecture, focusing on computer graphics, ray tracing, and sensor simulation, suitable for scenarios like digital twins, autonomous driving, and Siasun Robot & Automation, differentiating itself from H20. Regarding issues such as export controls, he stated the need to maintain good relationships with the governments of China, the U.S., and multiple countries worldwide, comply with policies, and address the impact of export control policies through enterprise flexibility. Noteworthy is that Huang Renxun has visited China three times by 2025, all due to invitations to events (such as the Chain Expo). He mentioned that his future frequent visits to China will depend on the invitations. He is optimistic about China's advantages in AI applications, supply chains, and electric vehicle fields, planning to deepen cooperation in China. In terms of NVIDIA's strategic layout and collaboration in China, Huang Renxun is optimistic about China's AI application innovation (including platforms like TikTok and Red), believing that China is leading in technology integration and application speed, with fierce market competition fostering high-quality capabilities that global companies are learning from. He pointed out that China's education system nurtures around 50% of global AI researchers, with a strong talent reserve in science, mathematics, and computer fields, creating a dynamic and active ecosystem. Therefore, he plans to expand talent recruitment in China, appreciating the professional abilities of Chinese engineers, some of whom have worked with NVIDIA for over 25 years. It is worth noting that NVIDIA's collaboration with Chinese companies focuses on technical support for areas like autonomous driving and foundational technologies required for large-scale model development, rather than directly participating in end-product manufacturing. Praising Huawei, DeepSeek, and Chinese AI innovation Right at the beginning, Huang Renxun was asked how he views the progress made by Chinese artificial intelligence and his thoughts on Huawei and Chinese companies. He clearly stated that "underestimating the capabilities of Huawei and Chinese manufacturing is extremely naive," highly recognizing Huawei's technological prowess, noting its awe-inspiring capabilities in high-end smartphone manufacturing, cellular telecommunications technology, etc. Although companies like Huawei and other Chinese firms have been around for a shorter time, they are competitive, with their ecosystem completeness only being a matter of time, and their involvement in technological competition itself showcases their strong capabilities. He believes that the Chinese AI industry can be divided into three layers: computer infrastructure, models, and applications, developing at a rapid pace; he affirmed the technological breakthroughs of model layer companies, such as DeepSeek launching the world's first open-source inference model, Alibaba's Thousand Questions, Kimi, and others being excellent technologies. At the same time, he disagreed with the viewpoint that "big models do not require massive AI research," acknowledging the innovative architecture of models like DeepSeek R1, believing they have made significant breakthroughs in inference efficiency. Finally, Huang Renxun mentioned AI. He believes that AI will revolutionize all industries and scientific fields, for example: AI driving the development of computer graphics (such as the graphics technology in "The Myth: Wu Kong"), supporting quantum computing pre-processing and post-processing through "CUDA Quantum" technology, assisting in physics simulations (such as large-scale weather forecasting), predicting biomolecular dynamics, Siasun Robot & Automation technology, etc. He views AI as the "greatest equalizer," capable of enhancing the capabilities of ordinary people (such as helping ordinary creators improve their skills) and accepting global scientific scrutiny through an open ecosystem to strengthen security. NVIDIA's competitive advantage lies in providing a general AI architecture and support throughout the lifecycle NVIDIA currently dominates the AI chip field, but competitors are quickly catching up. Companies like AMD, Huawei, Qualcomm, etc., all want a piece of the pie in this hot market. Regarding the competition from cloud vendors developing their own chips, Huang Renxun stated that NVIDIA's strategic difference from companies like Amazon, Google, etc., lies in focusing on providing a universal AI architecture and support throughout the lifecycle, rather than limiting themselves to developing their own chips. However, competition still has positive significance, driving industry progress, and the efforts of competitors prove the vitality of the market. Discussing NVIDIA's AI strategy and technological layout, he mentioned the need to build a whole lifecycle AI system, requiring a significant investment in the creation of an advanced system covering the entire AI lifecycle (pre-training, post-training optimization, reinforcement learning, inference). He also emphasized the openness and universality of AI architecture, supporting usage in various cloud platforms and enterprise-built systems, where users only need to understand NVIDIA's architecture and "language" to adapt flexibly. Currently, NVIDIA's strategic core goal is to maintain technological performance leadership, ensure system usability and practicality, and achieve a wide coverage of the architecture. Reflecting on NVIDIA's journey from being the "lowest global tech company by market value" to becoming the "highest tech company by market value," Huang Renxun emphasized their core contribution in "redefining computing" and pioneering the fundamental industry of AI. He believes that the U.S. computer industry (chips, systems, software, etc.) is a "national treasure," on par with China's supply chain and electric vehicle industry. NVIDIA's rapid growth is inseparable from talent, and Huang Renxun mentioned that the company adheres to the philosophy of "high pay incentivizing employees," considering NVIDIA as the "smallest large company globally" (with around 47,000 employees), with low turnover rates and teams that have worked together for a long time being a strength. This article is republished from CaiLian Press, GMTEight Editor: Chen Wenfang.