Here's an irony for the AI age: artificial intelligence has hit a bottleneck, and it's not what anyone expected. Microsoft (MSFT) has AI chips sitting around doing nothing because there's nowhere to plug them in. The constraint isn't silicon anymore. It's electricity.
Speaking on the "BG2Pod" technology and investing podcast hosted by Brad Gerstner and Bill Gurley, Microsoft CEO Satya Nadella laid out the problem alongside OpenAI CEO Sam Altman. Power shortages have become the industry's defining constraint, and it's changing how the entire sector operates.
When You Have the Chips But Not the Outlets
"The biggest issue we are now having is not a compute glut, but it's power," Nadella explained. "It's not a supply issue of chips. It's actually the fact that I don't have warm shells to plug into." By "warm shells," he means data centers that are either incomplete or lack the energy infrastructure and cooling capacity needed to run high-performance AI hardware.
This isn't theoretical. Microsoft has already slowed or paused some early-stage data center projects to deal with these constraints. In April, Microsoft Cloud Operations and Innovation President Noelle Walsh described the situation in a LinkedIn post as part of the "largest and most ambitious infrastructure scaling project" in the company's history.
Other tech giants are feeling the squeeze too. Alphabet Inc. (GOOG) has signed demand-response agreements with U.S. utilities, allowing Google to temporarily reduce data center power use during grid stress events and shift nonurgent computing tasks to off-peak hours. Amazon.com Inc. (AMZN) has outlined similar efforts to improve data center efficiency and support grid reliability as AI usage accelerates.
Altman's Warning on Energy Contracts
On the same podcast, Altman raised a different concern about energy markets. "If a very cheap form of energy comes online soon at mass scale, a lot of people are going to be extremely burned with existing contracts they've signed," he said. Companies locking in long-term power agreements today could find themselves stuck paying premium rates if energy costs suddenly drop.
Altman also pointed out that lower computing costs typically drive demand higher, which could strain infrastructure that's already stretched thin. The pace of AI growth, he argued, will ultimately depend on stable and affordable energy access. Sudden market shifts could disrupt operations across the industry.
Power Becomes the Competitive Advantage
Microsoft owns roughly 27% of OpenAI and holds exclusive Azure cloud rights to OpenAI's models through 2030. Both companies need reliable electricity to keep data centers running and expand capacity. In this environment, power access is becoming as strategic as computing power itself.
Energy producers have taken notice. Constellation Energy Corporation (CEG), Vistra Corp. (VST), and Brookfield Renewable Partners (BEP) have all signed major power deals with large technology firms, including arrangements for 24/7 carbon-free energy. These partnerships signal a fundamental shift in how tech infrastructure gets built and powered.
The International Energy Agency projects that global data center electricity consumption will roughly double to around 945 terawatt-hours by 2030. That would represent just under 3% of total global electricity demand. As AI workloads grow more intensive and widespread, the industry's appetite for power is only going to increase.
What started as a race for chips has turned into a scramble for kilowatts. The companies that solve the power problem first may end up with the biggest advantage in the AI economy.
