Nvidia CEO Jensen Huang laid out a bold forecast in an exclusive Bloomberg interview on July 27, 2026: the semiconductor industry must expand roughly tenfold over the next decade to serve a computing world dominated by autonomous machines rather than human users. Huang argued that Nvidia's addressable market is being redefined at the endpoint level, shifting from a billion people using computers to 100 billion agents and billions of robots all using computers.
Huang's thesis rests on the compute intensity of agentic AI versus generative AI. On Nvidia's May 2026 fiscal Q1 earnings call, he noted that reasoning models require 100 to 1,000 times more tokens per task than one-shot chatbots. That token demand compounds when agents run continuously without human rest cycles and spawn sub-agents. Huang emphasized that the semiconductor industry built on top of the chip industry 'is certainly not big enough' for this shift. He guessed the semiconductor industry would need to be 10 times larger over the next decade.
For infrastructure planners: Huang's 10x thesis directly implies CPU and accelerator capacity expansion, supporting South Korea's semiconductor ecosystem (which Huang flagged as key to the buildout) and raising questions about bottleneck timeframes for Nvidia's own supply chain. His comment that Nvidia sees 'bottlenecks everywhere' suggests demand visibility outpaces current manufacturing roadmaps, putting pricing and allocation pressure on endpoint deployments.