Anthropic Is Burning Gas to Feed the AI Boom—and It Says So Out Loud

Anthropic Is Burning Gas to Feed the AI Boom—and It Says So Out Loud


Anthropic is candid that its data centers will run on natural gas—even at a cost to the company's own stated values.


By Justine Calma


A Company Acknowledging the Trade-Off


The AI industry's appetite for computing power has collided with the limits of the power grid. As models grow larger and inference demands climb, hyperscalers and AI labs alike are scrambling for reliable, round-the-clock electricity. In 2026, that scramble increasingly points to a familiar but controversial answer: natural gas.


What sets Anthropic apart is not that it is turning to fossil fuels—many of its peers are—but that the company is openly saying so. Rather than burying the power source in sustainability footnotes, Anthropic has been clear that its data centers will run on gas, acknowledging the tension between its climate commitments and the energy realities of training and serving frontier models.


Why Gas, and Why Now


The logic is straightforward. AI data centers require firm, dispatchable power—electricity that is available every hour of the day, independent of whether the sun is shining or the wind is blowing. Battery storage and grid interconnection queues remain bottlenecked, and new nuclear capacity is still years from meaningful scale. Natural gas fills the gap: it is abundant, relatively cheap, and can be brought online faster than most alternatives.


For a company training and deploying large language models at scale, downtime is not an option. The economics of AI reward reliability, and gas plants—including the growing fleet of behind-the-meter installations built directly to serve data centers—deliver it.


The Reckoning with Stated Values


Anthropic has positioned itself as a safety- and responsibility-focused AI lab, and its public acknowledgment of gas-powered infrastructure reads as an attempt at honesty rather than marketing. The company is, in effect, admitting that the transition to clean energy is not moving fast enough to meet AI's demand curve.


That candor carries risk. Regulators, investors, and customers increasingly scrutinize the emissions footprint of AI services, and rivals may seize on the disclosure. Yet silence on power sources has become harder to sustain as energy consumption attracts mainstream attention.


The Broader Pattern


Anthropic's situation reflects an industry-wide tension. Data center electricity demand has surged, and utilities are revising forecasts upward year after year. Gas-fired generation is being revived and expanded to meet the load, while renewable procurement and storage build-outs race to catch up. The result is a period in which AI's growth and fossil fuel consumption are, for now, intertwined.


What to Watch


  • Disclosure and accountability: Will Anthropic and its peers publish detailed, verifiable data on energy sources and emissions?
  • Grid-scale solutions: How quickly can new nuclear, geothermal, and long-duration storage come online to displace gas?
  • Policy pressure: Will regulators require cleaner power for data centers, or will reliability concerns dominate?

For now, Anthropic's message is unusual in its clarity: the AI boom runs on gas, and the company knows it. Whether that honesty translates into faster action toward cleaner power is the question that will define the next phase of the industry's energy story.

via The Verge AI

Related