According to March 17, in the future, people may have to pay an extra utility bill:AI.
This is Sam Altman's point of view. He stated that artificial intelligence would eventually be traded as a basic utility, like electricity and water, and charged on a usage basis。
Local time last Wednesday, this one OpenAI At the Beled Infrastructure Summit in Washington, D.C., the CEO stated that a technology company like him was building a needs-based future for intelligence。
“Most fundamentally, our business, and I think all the other model providers' business, will end up looking like sales token,” Altmann says. Token here refers to the units of measurement used by artificial intelligence systems to process, price input and output data。
He added, “We see a future in which intelligence becomes a public utility like electricity or water, and people buy it through metering and use it for any scene they want to use.”
Altman states that in a world like this, arithmetic determines who can use artificial intelligence, and that the demand for artificial intelligence only increases. Calculus is the processing capacity required to train and run artificial intelligence models, determined by infrastructure such as chips, data centres, etc。
Ortman says if OpenAI The result is either “failure to provide services or extremely expensive”. He said that it would favour the use of artificial intelligence towards the rich or force the Government to decide how to allocate limited computing power。
To meet the soaring demand for artificial intelligence, major technology companies will invest hundreds of billions of dollars this year in the energy field。
In January this year, AMD Chief Executive Officer Susef, in his keynote speech at the International Consumer Electronics Fair (CES) 2026, stated that, over the next five years, the world would need more than 10 yottaflops (IT house note: 1 Yottaflops = 1024 FLOPS) computing power – a scale 10,000 times greater than that of 2022 global artificial intelligence。
Sustaining such expansion is a major infrastructure challenge. Artificial smart data centres can consume electricity more than small cities, while pressure on the United States grid, a shortage of transformers and slow approval of transmission lines can be development bottlenecks。
In January of this year, on a podcast entitled "Founding with Peter Demandes " , Elon Mask said that power generation is now a constraint to the development of artificial intelligence. He predicted that, given the faster pace of energy infrastructure development, China could outperform the United States in terms of artificial intelligence。
WITHIN SCIENCE AND TECHNOLOGY COMPANIES, NUMERACY IS A VALUABLE BUT SOMETIMES SCARCE RESOURCE. ENGINEERS COMPETE FOR ACCESS TO GRAPHIC PROCESSORS (GPUS), AND TODAY SOME JOB-SEEKERS ARE CONCERNED ABOUT THE ARTIFICIALLY INTELLIGENT BUDGET ALLOCATED TO THEM BY COMPANIES IN ADDITION TO ASKING ABOUT PAY AND EQUITY。
Last December, Greg Brockmann, President of OpenAI, stated that the company had committed about US$ 1.4 trillion over the next eight years to the data centre project in the hope that it would “be ahead of the trend”, but he also said, “I don't think we will be able to keep up, no matter how ambitious our dreams may be today.”
At the Beled Infrastructure Summit, Altmann stated that their goal was to get out of the “calculations constraints” situation。