NextFin News - OpenAI CEO Sam Altman appeared at BlackRock’s U.S. Infrastructure Summit in Washington, D.C., on March 11, 2026, joining a conversation hosted by Bayo Ogunlesi about AI growth, the infrastructure it requires, and public-private collaboration to accelerate delivery of power, data centers and the skilled trades needed to build them. (businesswire.com)
The discussion took place as part of BlackRock’s day-long summit on national infrastructure, convening public officials, corporate leaders and labor representatives to consider how the United States can scale capacity for emerging demands such as AI. Altman’s remarks focused on why AI is now as much an infrastructure problem as a technology one, how OpenAI is planning for abundant compute and energy, and what this means for workforce development, competition and governance. (businesswire.com)
Infrastructure scale and the cost of capacity
Altman opened by stressing the extraordinary capital intensity of modern AI: the infrastructure requirement is unprecedented and must be planned far in advance. He said that OpenAI spends heavily on infrastructure "in advance of revenue" and pursues business models that prioritize broad access over short-term profitability. In his words, OpenAI holds "a fundamental belief in abundance of intelligence," and aims "to flood the world with intelligence."
We want people to just use it for everything. We want this to just be something that the future generation doesn't think about.
Compute as the economic meter: tokens, pricing and utility
Altman described a commercial future where intelligence is metered like electricity or water: users will buy tokens to access models that vary in size, reasoning effort and cost. He argued that being capacity-constrained risks making access expensive or concentrated among the wealthy, and that the better path is to "flood the market" with compute to keep prices accessible.
If we don't have enough, we either can't sell it or the price gets really high and it kind of goes to rich people or society makes a bunch of sort of central planning decisions that I think almost always go badly.
Stargate, mega data centers and on-the-ground complexity
Altman described visiting large, gigawatt-scale campuses and the feeling of walking through vast halls under construction. He said OpenAI is training a leading model on its first site in Abilene and relayed the visceral shift from construction to operation — when "one researcher at OpenAI types in like one command and it pushes enter" and thousands of GPUs spin up. He acknowledged many expected and unexpected challenges: supply-chain issues, extreme weather events and the operational friction of building at scale.
Efficiency gains and model improvements
On model efficiency, Altman highlighted rapid progress: he estimated that moving from OpenAI’s early reasoning model (referred to as 01) to the integrated reasoning in model 5.4 yielded roughly a thousandfold reduction in the cost to reach the same answer on hard problems. He credited not only algorithmic advances but broader ingenuity across kernel engineers, power engineers and data-center designers that has driven energy and cost improvements.
From that first model to 5.4 has been a reduction in cost of about a thousandx.
Inference chip: an opinionated, energy‑constrained bet
Altman explained OpenAI’s decision to build a specialized inference-only chip. He framed the hardware as an "opinionated bet": not necessarily the fastest chip, but the most energy-efficient and cheapest for inference workloads, which will matter greatly in an energy-constrained world. He projected the first chips would be deployed at scale by the end of the year and that initial deliveries would arrive in a few months.
Skilled trades and workforce partnerships
Highlighting a partnership announced at the summit, Altman emphasized the need for many more skilled trades workers to construct and operate the physical infrastructure — power plants, transmission lines, data center halls and the mechanical systems inside them. He described the jobs as "incredible" and foundational for future economic prosperity and said OpenAI was working with the North America’s Building Trades Unions to expand training pathways. The summit itself emphasized workforce development as a core theme. (businesswire.com)
Competition with China and the global AI landscape
On global competition, Altman offered a nuanced reading: the U.S. leads on the most capable frontier models, while China is strong in cheaper inference usage and rapid industrialization and open‑source ecosystems. He warned that the U.S. is vulnerable if it falls behind on infrastructure, industrial capacity or economic adoption, and urged faster adoption across companies, scientists and government.
In terms of where we are, the most capable models in the world, the frontier the US is leading on. The cheapest inference usage for a two generation earlier model China is leading on.
Governance: democratic decisions over exceptional technology
Altman argued that because AI may reshape economies and geopolitics, decisions about its use and limits should be settled through democratic processes rather than by a handful of companies or by unilateral government edict. He said AI companies should have a meaningful voice, given their expertise, but the rules must be agreed upon by society; he called for a faster-moving democratic process to keep up with the technology’s pace.
I think this belongs to the will of the people working through the democratic process. Companies like ours... should have a real voice and we have opinions and understanding of where its limitations are and where it's not ready to be used.
Economic measurement, abundance and transition
Altman reflected on macroeconomic implications: AI could create an immense productivity boom and greatly improve quality of life, even while traditional measures such as GDP behave differently in an era of abundance. He urged policymakers and economists to consider new metrics and warned that the next few years will be a challenging adjustment period requiring intense debate and policy work.
Closing
The session closed with a forward-looking note: Altman emphasized both the practical constraints — power, supply chains, labor — and the potential for AI to become a ubiquitous utility. He urged rapid but careful action from industry, government and labor to build capacity and ensure broad access.
References
BlackRock: U.S. Infrastructure Summit press release (Business Wire), Feb. 10, 2026. (businesswire.com)
BlackRock LinkedIn announcement: Bayo Ogunlesi & Sam Altman conversation. (linkedin.com)
Explore more exclusive insights at nextfin.ai.
