FULL REMARKS: Sam Altman Addresses BlackRock U.S. Infrastructure Summit | AC15

AI Summary

Sam Altman, CEO of OpenAI, addressed the BlackRock U.S. Infrastructure Summit on March 11, 2026, asserting that AI has recently crossed a threshold into major economic utility, particularly in coding and knowledge work, with disorienting speed. He predicts a future where AI systems will proactively manage multi-day and multi-week tasks, functioning like senior employees with full context. Altman highlights a significant shift in startups prioritizing compute capacity over employee count, a trend he notes is slower for larger companies. He speculates that by late 2028, AI's cognitive capacity in data centers could surpass human cognitive capacity, and soon, major leaders will rely heavily on AI for their jobs. OpenAI recently secured a $110 billion funding round, four times larger than the largest public offering ever, to address the immense and expensive infrastructure needs, aiming to make intelligence "too cheap to meter." He envisions AI as a utility like electricity, with rapidly growing demand necessitating a "flood the market" approach to avoid high prices or central planning. Altman also discussed the Stargate project, with the first site in Abalene now training what he hopes will be the best model in the world, and expressed optimism about solving power demand challenges, citing a 1,000x cost reduction in reasoning models from 01 to 5.4. OpenAI is developing an inference-only chip focused on cost-efficiency for future agent demand and announced a partnership with North American building trades unions to expand training for skilled construction workers, crucial for building AI infrastructure. Altman views deep learning as a scientific discovery, not a trade secret, and believes the US leads in frontier models and infrastructure, while China leads in cheaper inference usage and open-source AI. He was particularly impressed by the aggressive adoption of AI by startups and large companies in India. Finally, Altman emphasizes that AI's societal impact necessitates democratic decision-making, not just corporate or governmental control, and warns of US vulnerabilities in global supply chains, slow economic adoption, and potential policies hindering global diffusion of American AI tech. He foresees an immense productivity boom but also a painful adjustment period as society shifts from managing scarcity to abundance.

Claims Extracted (13)