In a wide-ranging discussion on the No Priors Podcast, NVIDIA Founder and CEO Jensen Huang reflects on the rapid evolution of artificial intelligence throughout 2025 and provides a strategic roadmap for 2026. From the debunking of the “AI Bubble” to the rise of physical robotics and the “ChatGPT moments” coming for digital biology, Huang offers a masterclass in how accelerated computing is reshaping the global economy.

---

TL;DW (Too Long; Didn’t Watch)

• The Core Shift: General-purpose computing (CPUs) has hit a wall; the world is moving permanently to accelerated computing.
• The Jobs Narrative: AI automates tasks, not purposes. It is solving labor shortages in manufacturing and nursing rather than causing mass unemployment.
• The 2026 Breakthrough: Digital biology and physical robotics are slated for their “ChatGPT moment” this year.
• Geopolitics: A nuanced, constructive relationship with China is essential, and open source is the “innovation flywheel” that keeps the U.S. competitive.

---

Key Takeaways

• Scaling Laws & Reasoning: 2025 proved that scaling compute still translates directly to intelligence, specifically through massive improvements in reasoning, grounding, and the elimination of hallucinations.
• The End of “God AI”: Huang dismisses the myth of a monolithic “God AI.” Instead, the future is a diverse ecosystem of specialized models for biology, physics, coding, and more.
• Energy as Infrastructure: AI data centers are “AI Factories.” Without a massive expansion in energy (including natural gas and nuclear), the next industrial revolution cannot happen.
• Tokenomics: The cost of AI inference dropped 100x in 2024 and could drop a billion times over the next decade, making intelligence a near-free commodity.
• DeepSeek’s Impact: Open-source contributions from China, like DeepSeek, are significantly benefiting American startups and researchers, proving the value of a global open-source ecosystem.

---

Detailed Summary

The “Five-Layer Cake” of AI

Huang explains AI not as a single app, but as a technology stack: Energy → Chips → Infrastructure → Models → Applications. He emphasizes that while the public focuses on chatbots, the real revolution is happening in “non-English” languages, such as the languages of proteins, chemicals, and physical movement.

Task vs. Purpose: The Future of Labor

Addressing the fear of job loss, Huang uses the “Radiologist Paradox.” While AI now powers nearly 100% of radiology applications, the number of radiologists has actually increased. Why? Because AI handles the task (scanning images), allowing the human to focus on the purpose (diagnosis and research). This same framework applies to software engineers: their purpose is solving problems, not just writing syntax.

Robotics and Physical AI

Huang is incredibly optimistic about robotics. He predicts a future where “everything that moves will be robotic.” By applying reasoning models to physical machines, we are moving from “digital rails” (pre-programmed paths) to autonomous agents that can navigate unknown environments. He foresees a trillion-dollar repair and maintenance industry emerging to support the billions of robots that will eventually inhabit our world.

The “Bubble” Debate

Is there an AI bubble? Huang argues “No.” He points to the desperate, unsatisfied demand for compute capacity across every industry. He notes that if chatbots disappeared tomorrow, NVIDIA would still thrive because the fundamental architecture of the world’s $100 trillion GDP is shifting from CPUs to GPUs to stay productive.

---

Analysis & Thoughts

Jensen Huang’s perspective is distinct because he views AI through the lens of industrial production. By calling data centers “factories” and tokens “output,” he strips away the “magic” of AI and reveals it as a standard industrial revolution—one that requires power, raw materials (data/chips), and specialized labor.

His defense of Open Source is perhaps the most critical takeaway for policymakers. By arguing that open source prevents “suffocation” for startups and 100-year-old industrial companies, he positions transparency as a national security asset rather than a liability. As we head into 2026, the focus is clearly shifting from “Can the model talk?” to “Can the model build a protein or drive a truck?”