Anthropic recently completed the largest initial public offering in history, raising approximately $81 billion and cementing its status as a dominant force in the artificial intelligence sector. While the financial milestone underscores the deep integration of AI into the global economy, the company’s pre-IPO regulatory filings reveal a complex landscape regarding energy consumption and climate strategy.
Previously, Anthropic relied on leasing data center space from major cloud providers such as Amazon and Google. However, intensifying competition has driven the company to commission its own dedicated data center infrastructure and power supplies. This shift has led to increased reliance on natural gas to meet substantial power demands, a trend noted by other AI firms seeking to scale operations.
The reliance on fossil fuels is part of a broader trend in the sector. The International Energy Agency’s latest World Energy Outlook identified 180 gigawatts of new gas power capacity planned for U.S. data centers, with a significant portion expected to come online during the 2030s. This volume represents roughly 14% of the current total utility-scale generation capacity.
Despite its reputation as a public-benefit corporation founded on the principle of using AI responsibly, Anthropic’s filing offers limited detail on its emissions profile. While the company outlines mechanisms to reduce grid strain and addresses regulatory pressures regarding electricity prices, it frames the risks associated with natural gas primarily in terms of cost and availability rather than environmental impact. The filing acknowledges natural gas will play a role but does not provide extensive insight into how the company plans to mitigate the climate consequences of its infrastructure.
Industry analysts note that while public attention currently focuses on energy affordability, the long-term commitment to gas power could complicate Anthropic’s social license to operate, particularly in jurisdictions like Europe where emissions disclosures are increasingly stringent. As competition in AI compute intensifies, companies may need to address climate considerations more directly to maintain their competitive advantage and public standing.
If they cannot manage their own carbon footprint, why does anyone trust them to be responsible AI leaders? Doubtful.
Honestly, I am not surprised. Training those massive models consumes absurd amounts of power, so gas is just the cheap shortcut.
It is wild that the biggest tech IPO is secretly relying on natural gas. What happened to the green promises?