Google’s ex-CEO Eric Schmidt has raised an alarm over America’s energy capacity. In a video shared on social media platform X (formerly known as Twitter), Schmidt said that America needs an additional 92 gigawatts of power to sustain the rapid growth of artificial intelligence. In the video, Schmidt has cautioned that the US is ‘running out of electricity’, highlighting the strain posed by massive data centers, intensive cooling requirements and round-the-clock operations.

Increasing energy concerns

Schmidt’s remarks echo earlier warnings from tech leaders: Chamath Palihapitiya predicted electricity rates could double within five years without structural reforms. On the other hand, Jeff Bezos suggested orbital data centers could rival terrestrial facilities within two decades, leveraging uninterrupted solar power in space.

Google’s Space Data Center Moonshot

Alphabet CEO Sundar Pichai reaffirmed Project Suncatcher in December 2025, a plan to test orbital data center prototypes by 2027. The initiative aims to harness solar power in space and reduce cooling constraints. Pichai described it as a “moonshot,” comparable to Waymo, and credited SpaceX’s breakthroughs in launch technology for making the effort feasible.

Microsoft, Anthropic and OpenAI’s electricity pledge

Recently, three of the world’s biggest AI companies made nearly identical promises within 30 days. Microsoft kicked things off on January 13, pledging to cover the full electricity costs of its data centres. OpenAI followed on January 21. Anthropic joined on February 11. Same playbook. Same month.The numbers behind this rush are stark. Lawrence Berkeley National Lab found data centres consumed 4.4% of US electricity in 2024 and could hit 12% by 2028. Carnegie Mellon researchers estimated this demand surge could push electricity generation prices up by 25% in data centre-heavy markets by 2030. Consumers on the largest US grid, PJM Interconnection, face a $16.6 billion bill from 2025 through 2027 just to secure future power supply for these facilities.That kind of math turns into a political problem fast. And it already has.