In the future, we may have to pay another utility bill: artificial intelligence.
This is because, according to Sam Altman, AI will eventually be bought and sold as basic utilities like electricity and water that are metered according to usage.
Speaking Wednesday at the BlackRock Infrastructure Summit in Washington, D.C., OpenAI’s CEO said technology companies like his are building for a future where intelligence is delivered on demand.
“I think basically our business, and every other model provider’s business, is going to be like selling tokens,” Altman said, referring to the units that AI systems use to process and price input and output data.
“We are seeing a future where intelligence is a utility like electricity or water, and people can buy it at the meter and use it for whatever they want,” he added.
In that world, computing power will determine who has access, and the demand for AI will only increase, Altman says. Computing power is the processing power needed to train and run AI models and is determined by infrastructure such as chips and data centers.
If OpenAI can’t build enough computing power to meet demand, Altman said, “either we can’t sell it or the price will be very high.” That could skew access to AI to the wealthy, he said, or force governments to decide how much limited computing should be distributed.
infrastructure sprint
Big tech companies plan to spend hundreds of billions of dollars on computing this year to meet the surge in demand for AI.
AMD CEO Lisa Su said in her keynote at CES 2026 in January that the world will need more than “10 Yottaflops” of computing over the next five years to keep up with growth (10,000 times the world’s AI capacity in 2022).
Driving that expansion is a major infrastructure challenge.
AI data centers can consume as much electricity as a small city, and the strain on the U.S. power grid, along with transformer shortages and transmission line permitting delays, could become a bottleneck.
Elon Musk said in a January episode of his Moonshot with Peter Diamandis podcast that power generation is now the limiting factor in AI expansion, and predicted that China could surpass the United States in total AI computing power because it ramps up energy so quickly.
Inside technology companies, computing is a valuable resource, but sometimes in short supply. Engineers are competing for access to GPUs, and some job seekers are asking about budgets for AI computing alongside salary and capital.
Last December, OpenAI President Greg Brockman said the company, which has committed about $1.4 trillion to data center projects over the next eight years, wants to “stay ahead of the curve,” but “no matter how wildly we dream at this point, I don’t think we’ll ever get there.”
At the BlackRock Infrastructure Summit, Altman said the goal is to move away from a “capacity-constrained” world.
