The scale of the shift is stark. In 2019, Amazon promised carbon neutrality by 2040, while Google and Microsoft targeted 2030 for carbon-free operations. Today, those goals are clashing with the massive energy requirements of AI development. Amazon is currently developing a gas-fired power plant in Texas that, according to regulatory records, could emit 33 million tons of carbon dioxide annually—potentially becoming the largest single source of power-related emissions in the United States.

Despite the optics, Amazon ranks as the cleanest among major players according to Climate Power’s Data Center Air Pollution Tracker, scoring 68 out of 100. The index paints a far darker picture for Elon Musk’s xAI. With a score of just 6, xAI relies entirely on behind-the-meter gas power, producing 898 pounds of carbon dioxide per megawatt-hour. This makes the firm nearly ten times dirtier than Oracle, which sits at 57 points.

However, these scores do not account for total volume. The collective carbon footprint of Amazon, Microsoft, and Google reached 19 million metric tonnes of carbon dioxide equivalent last year—roughly one-third of France’s total emissions. Experts suggest these figures likely underestimate the true environmental cost. Cecilia Rikap, an economics professor at University College London, argues that cloud giants are essentially obscuring the carbon footprints of other corporations, which outsource their data storage and AI training to these platforms. As these companies continue to scale, their public commitment to sustainability faces increasing scrutiny from regulators and academics alike.