Exclusive Interview with NVIDIA CEO Jensen Huang: Debunking AI Doomsday Theories, Optimistic About the Future of Open Models and Computing Infrastructure in the Next Decade

This episode of "Jensen Huang says the AI doomers have it wrong" (Axios interview, conducted by Mike Allen) summarizes the core content as follows: 1. Geopolitics, Export Controls, and Open Source Ecosystem • Views on US-China geopolitical competition: hopes for open technology and research exchanges. Half of the world's AI researchers are in China, contributing groundbreaking research. • Core value of open models: • The market needs not only closed-source services like Anthropic and OpenAI but also open models. • Openness is the foundation of scientific progress, cybersecurity, national security, and economic security. Through open and transparent oversight and testing, the community can better strengthen security defenses (cybersecurity requires distributed self-defense). • Use of Chinese open models (such as Kimi, DeepSeek, Qwen, etc.): • US companies should be allowed to use excellent Chinese open models, with no so-called "backdoor" risks. Models typically run in isolated and secure "sandboxes" and "harnesses." • Emphasizes that excellent open models will promote the entire industry's use of AI, and increased usage means a greater need for NVIDIA's computing infrastructure and data centers, which is a significant benefit for the entire industry and hardware suppliers. 2. Refuting "AI Doomsday" and "Job Loss Panic" • Criticism of "AI doomsday theories": • Believes that some tech leaders' claims that "AI will destroy humanity" or "will eliminate half of US jobs" are baseless fantasies and hype. • This fictional doomsday narrative only creates unnecessary social panic and hinders the rapid application of AI in the industry. • AI creates jobs rather than destroys them: • Historical and current facts prove that productivity improvements lead to more industry opportunities. • Automation increases efficiency (e.g., radiologists automatically reviewing scans, paralegals automatically processing documents), significantly enhancing service capacity and increasing the number of patients/customers, which in turn drives further demand for these positions. • The current AI boom has directly driven an explosion of jobs in manufacturing, energy, chips, and infrastructure sectors. • The distinction between jobs and tasks: AI automates specific "tasks," not the ultimate "purpose" of a profession. For example, typing is just a task; solving problems and creating value are the essence of work itself. 3. NVIDIA's Product Layout and Market Forecast • NVIDIA Nemotron open model: not aimed at competing with closed-source giants in the general large model space, but rather providing customizable open foundational models for enterprises needing sovereignty, privacy protection, and domain-specific knowledge. • Computing expenditure and "bubble theory": • Denies the danger of "AI computing surplus/bubble" in the short term. AI is currently in the earliest stage of infrastructure development (unlikely to burst in the next 5-10 years). • The current bottleneck lies in physical constraints (chip, memory, power, land, and a shortage of data center construction workers), which can actually prevent overheating. • Long-cycle transformation of the semiconductor industry: • This round of prosperity is not driven by cyclical or seasonal consumer demand but by a historic reconstruction of computer infrastructure (from pre-recorded static content to real-time generated intelligent tokens). • Predicts that the semiconductor industry needs to expand 5 to 10 times in scale within the next 10 years to meet demand. 4. Evolution of AI and the Era of Robotics • The essence and value of tokens: Tokens are mathematical units embedded with knowledge and intelligence. As models become smarter, the value contained in each token increases, leading customers to be more willing to pay, forming a commercial flywheel. • The era of agentic AI: In the future, there will be hundreds of billions or even trillions of AI agents running in the background and collaborating globally, leading to explosive demand for computing infrastructure geometrically. • The "ChatGPT moment" for robots: The "ChatGPT moment" for robots has arrived (showing surprising reasoning and spatial understanding abilities, such as autonomously understanding to open the cabinet door before placing the apple); it is expected that within the next 3 to 4 years, robots will truly enter a stage of widespread and practical use. 5. Leadership and Personal Philosophy • Pain and tempering: Excellence comes from "a lot of pain and suffering" and repeated practice when no one is watching. These setbacks shape character, resilience, and a calm mindset when facing high pressure. • Confidence in America: Still believes in the "American Dream." The US has an open discussion environment, a soil for free innovation and collaboration, and as long as it continues to embrace and apply new technologies, it can maintain its lead in the next industrial revolution. • Time management philosophy: Does not wear a watch or strictly constrain himself with Outlook, maintaining a focus on "the present moment (Now is the most important time)" and being highly focused on spending time with family and pets on weekends.