Sam Altman, CEO of OpenAI, recently raised concerns about the impending challenges surrounding power and data center infrastructure necessary for the rapidly evolving artificial intelligence (AI) landscape. Through the podcast, he focused the discussion on the cataclysmic problem of keeping up with exploding demand for computing power. This demand has outstripped the capacity of traditional utility providers.
Altman has gone all in on his clean energy investments, such as nuclear and solar-related initiatives. The point that really struck me was that as a software-first business, OpenAI is really suffering from infrastructure growing pains. He lamented that costs tied to expansive data center build-out are proving more troublesome.
Alongside Altman’s perceptions, Microsoft CEO Satya Nadella expressed the same lack of creative freedom that they’ve been limited to. He helped define the challenge. It’s not just a chip shortage—far from it—it’s a deficit of “warm shells,” a term for the boilerplate infrastructure necessary to deploy them.
“If you can’t do that, you may actually have a bunch of chips sitting in inventory that I can’t plug in. In fact, that is my problem today. It’s not a supply issue of chips, it’s the fact that I don’t have warm shells to plug into,” – Satya Nadella on the BG2 podcast.
This dynamic is emblematic of a larger trend taking place in the tech industry. Electricity demand across the United States has been flat, or even in decline, for more than a decade. Recent growth trends suggest an overwhelming tidal wave of new demand from data centers. This increase has exceeded utilities’ expectations for new generating capacity additions by five years and counting. The market for new gas turbines won’t pick up again until later this decade. This will make it more difficult to replicate the success of expanding infrastructure.
Altman further explained why this development is so significant during the short podcast episode. He flagged the challenge of meeting exponentially growing demand given new advancements in AI technologies.
“If we can continue this unbelievable reduction in cost per unit of intelligence — let’s say it’s been averaging like 40x for a given level per year — you know, that’s like a very scary exponent from an infrastructure buildout standpoint,” – Sam Altman.
Altman offered up a dire prediction of what would happen if cheaper forms of energy were identified and deployed widely. He pointed out that previously negotiated contracts may become anchoring weights as energy companies dash to respond to the new competitive energy market.
“If a very cheap form of energy comes online soon at mass scale, then a lot of people are going to be extremely burned with existing contracts they’ve signed,” – Sam Altman.
TechCrunch’s senior climate reporter Tim De Chant has been watching these machinations closely. His years of experience in environmental science and policy certainly adds muscle to Altman and Nadella’s worries. They’re concerned about how sustainable it is to power this new generation of AI technologies.
The relationship between energy efficiency and demand is equally impressive. Contrary to this intriguing attitude, Altman seems to embrace Jevons Paradox. This popular economic theory posits that when technology makes industries more resource efficient, overall consumption actually increases rather than decreases.
“If the price of compute per like unit of intelligence or whatever — however you want to think about it — fell by a factor of 100 tomorrow, you would see usage go up by much more than 100 and there’d be a lot of things that people would love to do with that compute that just make no economic sense at the current cost,” – Sam Altman.
The need for answers is clearly felt by all disruptive innovators just like industry titans, Altman and Nadella as they find their way through these choppy waters. While demand for data centers is booming, our energy infrastructure is unable to meet that unprecedented demand. We’ve got to address these issues to best position ourselves to enable smart and sustainable growth in AI tech.