Jensen Huang Claims AGI Is Already Achieved, Ilya Is Wrong, Programmers to Reach 1 B

During a two‑hour conversation with Lex Fridman, Nvidia CEO Jensen Huang made a series of bold statements about the current state and future trajectory of artificial intelligence. He began by declaring,

We have already achieved AGI.

He argued that the pre‑training scaling law still holds, emphasizing that data quantity—not model size—remains the primary limiter of AI intelligence, and that synthetic data will dramatically expand the data pool.

Huang directly challenged Ilya Sutskever’s claim that the data supply is exhausted, stating that the industry’s panic is unfounded. He explained that most data exchanged in research is already synthetic, generated, and re‑processed by humans, and that post‑training will increasingly rely on synthetic data while the bottleneck shifts from data to compute power.

Regarding inference, Huang disputed the notion that inference is lightweight, describing it as a form of thinking that is far more complex than simple reading. He highlighted that inference will become a massive market for specialized chips, but the real value lies in the ability of AI agents to plan, search, and act during test time.

Huang introduced the concept of “agentic scaling,” where agents generate new agents, creating a multiplicative effect on AI capabilities. He illustrated this with the OpenClaw platform, calling it the "iPhone of the token era," and explained how token‑priced AI services will be layered like iPhone tiers, with free, premium, and intermediate tokens.

On data‑center infrastructure, Huang discussed the mismatch between grid capacity and actual usage, noting that 99% of the time the grid is under‑utilized. He proposed designing data centers that can gracefully degrade performance, shift workloads, or use backup generators during peak grid demand, thereby reducing reliance on costly grid expansions.

Huang also reflected on Nvidia’s supply‑chain strategy, praising TSMC’s ability to coordinate global demand and emphasizing the importance of trust, technical excellence, and customer service in maintaining high yield and capacity.

When asked about management philosophy, Huang compared building a company to designing a computer system: start with a clear specification of the desired output, then iterate. He described his leadership style as highly collaborative, holding large meetings with all senior staff, encouraging open debate, and making decisions visible to the entire organization. He emphasized curiosity‑driven reasoning, rapid iteration, and the need to constantly share the reasoning process with the team.

Huang addressed concerns about AI‑driven job displacement, arguing that while specific tasks may be automated, the core purpose of professions—such as radiologists or programmers—remains unchanged and often expands. He predicted that the number of people capable of programming could grow from 30 million to a billion as programming becomes more about specification than code writing.

Finally, Huang expressed optimism about humanity’s future, citing confidence in human kindness, generosity, and the ability to solve problems. He envisioned a world where AI becomes a commodity, enabling breakthroughs in healthcare, environmental sustainability, and even space travel, while urging individuals to learn AI tools early and embrace the transformative potential of the technology.