Nvidia CEO Jensen Huang said AI infrastructure investment remains in an acceleration phase, with supply capacity rather than demand limiting growth. He reiterated that the global AI infrastructure market could reach $3 trillion to $4 trillion by 2030 and said Nvidia's fiscal 2027 revenue growth expectation of roughly 70% reflects a supply ceiling. Huang described the investment cycle as a shift from retrieval-based computing to generative computing, in which systems produce responses in real time and require a continuously operating computing layer. As Moore's Law delivers diminishing gains at the individual-chip level, the industry is turning to system-scale architectures involving GPU clusters, CoWoS advanced packaging (chip packaging for high-bandwidth integration) and NVLink interconnects. Huang also defended Nvidia's investments in AI companies and compute providers, saying investees have approximately $100 billion in customer contracts and revenue streams. The conference highlighted cybersecurity as an emerging AI market, persistent shortages across semiconductor equipment, storage and cleanroom capacity, SpaceX's expansion into integrated AI infrastructure, and an ongoing debate over whether the spending cycle is a supercycle or a bubble. Goldman Sachs researchers said the next test is whether hyperscale providers can convert capital expenditure and backlogs into durable enterprise revenue, margins and cash flow.