Back to articles

AI data center growth prompts Meta to revisit its 2030 net-zero goal

3 min read

Meta is reconsidering the timetable for reaching net-zero emissions by 2030. In its 2026 sustainability report, the company said its long-term ambitions remain unchanged, but that demand for AI infrastructure has grown faster than expected. It is therefore reviewing the timing, sequence, and interim milestones of its climate plans. This is an acknowledgment of pressure on the original schedule—not an announcement that Meta has abandoned its net-zero commitment or set a replacement date.

Emissions rose faster than at major tech peers

Meta reported total greenhouse gas emissions of more than 21.7 million metric tonnes of carbon dioxide equivalent in 2025, up from 15.6 million tonnes in 2024. The increase was nearly 40 percent. The company attributed much of the growth to “capital goods,” including data center servers and other equipment. Based on the figures cited in the report, Meta’s year-over-year increase exceeded Microsoft’s 25 percent, Google’s 18 percent, and Amazon’s 16 percent. A faster rate of growth does not mean Meta has the largest overall footprint among the major tech companies; the article notes that its total may still be comparatively small.

The company also accounts for measures such as renewable energy projects and carbon removal. Even after those efforts are factored in, however, Meta’s reported net emissions rose from more than 8.1 million tonnes in 2024 to more than 11.8 million in 2025—a 45 percent increase. The data show that these measures did not offset the year’s emissions growth. The report’s broader context is an industry building data centers at speed, with some projects still relying on natural-gas generation. That leaves companies balancing the need for computing capacity against their climate commitments.

Water figures tell a more complicated story

Water used directly at Meta’s facilities declined from 825 million gallons in 2024 to 790 million in 2025. But water use in areas experiencing water stress increased from 198 million to 246 million gallons. The report also tracks water associated with purchased electricity: the water consumed in generating power for Meta rose from 19 billion gallons to more than 26 billion gallons. That indirect figure highlights how a data center’s water footprint can extend well beyond cooling systems at the site itself.

Meta’s review is therefore about more than whether a corporate deadline will move. It points to the difficulty of assessing AI’s environmental costs when demand, server purchases, power generation, and water use are intertwined. Reporting only direct facility consumption would leave out a substantial part of the picture. Ars Technica notes that Meta is the only hyperscaler to report this indirect water use. As data center construction accelerates, the clarity of corporate targets—and the scope and consistency of environmental disclosures—will shape how the public can assess progress across the industry.

Ars Technica AI

Comments

Checking sign-in status...

Loading comments...

Related articles