Huang predicts 5-10x chip industry growth as AI agents and robots reshape computing demand

Nvidia's Jensen Huang recently outlined a vision of massive semiconductor growth, projecting that the industry must expand five to tenfold over the coming decade to support the rise of autonomous software agents and robotics. Rather than the current paradigm where roughly a billion people operate computers, Huang envisions a future in which 100 billion agents and billions of robots run continuously on computational infrastructure.

From human users to autonomous systems#

The shift represents a fundamental recalculation of what "computing demand" means. Instead of peak usage tied to human work hours or leisure patterns, AI agents and robots would operate around the clock, generating persistent, background computational load. Huang has been laying groundwork for this narrative for months, including recent statements positioning traditional software engineering as obsolete in favor of building AI agent systems.

On Nvidia's earnings call in May, Huang characterized the transition from generative AI to agentic AI—systems capable of reasoning, planning, and acting autonomously—as the industry's next transformative phase.

Infrastructure strain at scale#

This shift creates ripple effects across backend systems. As API traffic increasingly originates from autonomous agents rather than sporadic human interactions, long-standing assumptions about rate limiting, session memory, latency routing, and gateway capacity begin to fail. Backend infrastructure must be redesigned from the ground up to handle continuous, high-volume requests from non-human endpoints rather than bursty human-driven patterns.

To support this explosion at scale, the physical supply chain must grow dramatically, particularly in memory production and data center deployment. Huang recently highlighted South Korea's role as critical infrastructure, calling it "the beginning of a golden age," citing the nation's semiconductor and manufacturing strengths as essential to global AI buildout.

Major partnerships and locked-in supply#

Nvidia announced a partnership with SK Group valued at over $500 billion, covering bulk purchasing of next-generation High-Bandwidth Memory from SK Hynix, joint development of custom HBM4 chips optimized for agentic workloads, and supercomputer deployment. South Korean companies are moving aggressively: SK Telecom plans a 2-gigawatt data center using Nvidia's Vera Rubin architecture, expected online in 2027, while Nvidia will invest $1 billion in Naver Corp alongside Brookfield's up-to-$9-billion commitment to expand its AI data center capacity from 55 megawatts to 200 megawatts by 2028.