OpenAI's Sam Altman says AI query energy use fell up to 100,000x at ADIPEC
In brief
- Per-query AI energy use fell 1,000x to 100,000x in three to four years, Altman estimated.
- ADIPEC energy conference in Abu Dhabi hosted his remarks on November 2, 2026.
- No methodology accompanied the estimate.
- Energy is the AI industry's primary challenge, Altman said.
- OpenAI, ADNOC and G42 entered a multi-year partnership, Altman also announced.
A hedged estimate, not a measurement
Altman credited the drop in per-query energy to the development of more efficient AI models. That's the explanation he gave.
The range itself is wide (two orders of magnitude separate the low end from the high end), and it's Altman's own figure. Crypto Briefing noted that no breakdown of how it was calculated accompanied the remarks, so it isn't measured data and shouldn't be read as such.
"thousands or hundreds of thousands of times maybe more efficient"
That's how Altman described today's queries compared with those of a few years ago.
The "maybe" matters.
Energy is still the bottleneck
Altman didn't present the efficiency gains as a solved problem. He told the audience that energy is the primary challenge facing the AI industry.
Demand is part of the picture. Crypto Briefing, citing a research summary of his remarks, reported that top AI users now consume hundreds of billions of tokens every month, against roughly 100,000 tokens six years ago. The report didn't specify where that summary came from.
Altman has repeatedly advocated for expanding energy generation capacity, pointing to renewables and nuclear power. He's also pushed for what he describes as a more intelligent conversation about AI's energy needs, and he expressed hope that AI could eventually improve global energy availability.
OpenAI, ADNOC and G42
Altman used the same stage to announce a multi-year partnership between OpenAI, the Abu Dhabi National Oil Company (ADNOC) and Abu Dhabi-based AI firm G42, as reported by Crypto Briefing. The deal aims to deploy frontier AI models across ADNOC's operations, with the goal of optimizing how the energy company runs its business.
It's a notable pairing: an AI lab that calls energy its biggest constraint, signing on with one of the region's largest energy producers at an energy conference.
Frequently asked questions
How much less energy do AI queries use, according to Sam Altman?
Altman estimated that AI queries use between 1,000 and 100,000 times less energy than they did three to four years earlier. He credited the drop to more efficient AI models. The range is his own hedged estimate, and no breakdown of how it was calculated accompanied the remarks.
Why does Altman still call energy AI's main challenge?
Altman told the ADIPEC audience that energy is the primary challenge facing the AI industry. Crypto Briefing, citing a research summary of his remarks, reported that top AI users now consume hundreds of billions of tokens monthly, compared with roughly 100,000 tokens six years ago.
