Jensen Huang Shares His Vision For NVIDIA, Also Deepening His Commitment to China’s AI Future

Following his participation in the 3rd China International Supply Chain Expo (CISCE), NVIDIA CEO Jensen Huang attended a media exchange session. During the more than one-hour discussion, Huang addressed key topics including NVIDIA’s technology strategy, industry competition, China market positioning, and AI development, demonstrating NVIDIA’s confidence in China’s AI prospects and its commitment to the Chinese market.

New RTX Pro GPU Targets Digital Twins, Autonomous Driving, and Robotics

Huang emphasized NVIDIA’s focus on the Chinese market, expressing hopes to provide China with chips more advanced than the H20. While the H20 still holds advantages in inference capabilities and bandwidth due to its Hopper architecture, Huang noted that technology continues to evolve, and future products permitted for export to China will see upgrades. He also mentioned that the resumption of H20 sales in China depends on U.S. government licensing approval, and while supply chain reactivation will take time, customer demand remains strong.

Huang introduced the newly released RTX Pro GPU, based on the Blackwell architecture, which focuses on computer graphics, ray tracing, and sensor simulation. It is designed for applications such as digital twins, autonomous driving, and robotics, differentiating it from the H20.

Regarding export controls, Huang stated that NVIDIA must maintain good relations with governments in China, the U.S., and other countries, complying with policies while adapting to regulatory impacts through corporate flexibility.

Notably, this marks Huang’s third visit to China in 2025, all due to invitations. He indicated that future visits would depend on invitations but reiterated his optimism about China’s strengths in AI applications, supply chains, and electric vehicles, pledging to deepen cooperation in China.

NVIDIA’s China Strategy: AI Talent, Ecosystem, and Partnerships

Huang praised China’s AI application innovation (including platforms like TikTok and Xiaohongshu), noting that China leads in rapid technology integration and application. He highlighted the fierce market competition that has cultivated high-quality capabilities, with global companies learning from China’s ecosystem. He pointed out that China’s education system produces about 50% of the world’s AI researchers, with strong talent reserves in science, math, and computer fields, fostering a dynamic innovation environment.

As a result, NVIDIA plans to expand hiring in China, commending the expertise of Chinese engineers, some of whom have worked with him for over 25 years.

NVIDIA’s collaboration with Chinese companies focuses on technical support (e.g., autonomous driving, foundational tech for large models) rather than direct involvement in end-product manufacturing.

Praising Huawei, DeepSeek, and China’s AI Innovation

When asked about China’s AI advancements, Huang thinks underestimating Huawei and China’s manufacturing capabilities is extremely naive, acknowledging Huawei’s formidable strength in high-end smartphones and telecom technologies. He noted that while Chinese companies like Huawei have developed rapidly, their competitiveness is undeniable, and ecosystem maturity is only a matter of time.

He noted that China's AI industry can be divided into a three-layer architecture: computing infrastructure, models, and applications, all developing at an extremely rapid pace. He specifically acknowledged technological breakthroughs at the model layer, citing examples like DeepSeek's launch of the world's first open-source inference model, while also recognizing Alibaba's Tongyi Qianwen and Kimi as outstanding technologies.

He also praised the innovative architecture of models like DeepSeek R1, highlighting their significant breakthroughs in inference efficiency.

NVIDIA’s Competitive Edge: Universal AI Architecture & Full-Lifecycle Support

Despite NVIDIA’s dominance in AI chips, competitors like AMD, Huawei, and Qualcomm are rapidly catching up.

Addressing competition from cloud providers (e.g., Amazon, Google) developing in-house chips, Huang said NVIDIA’s advantage lies in offering universal AI architecture and full lifecycle support, rather than limiting itself to proprietary chips. He welcomed competition as a driver of industry progress, noting that rivals’ efforts validate market vitality.

When discussing NVIDIA's AI strategy and technological roadmap, Huang emphasized the company's commitment to building full-lifecycle AI systems. This requires substantial investment to develop advanced systems covering the entire AI lifecycle - including pre-training, post-training optimization, reinforcement learning, and inference.

He particularly stressed the importance of open and universal AI architectures that can be deployed across various cloud platforms and enterprise-built systems. Users only need to master NVIDIA's architecture and "language" to achieve flexible adaptation.

Currently, NVIDIA's strategic core objectives focus on three key areas: maintaining technological performance leadership, ensuring system usability and practicality, and achieving widespread architectural adoption.

Reflecting on NVIDIA's journey from being the "lowest market-cap tech company globally" to becoming the "highest market-cap tech company," Huang highlighted that its fundamental contribution lies in "redefining computing" and pioneering AI as a foundational industry.