--- title: "Microsoft's Carbon Emissions Jumped 25% as AI Data Centers Boomed" description: "Microsoft's carbon emissions rose about 25% last year, its sustainability report shows, as a frenzy of AI data-center building drove up the concrete, steel and power behind its cloud. The jump pushes the company further from its 2020 pledge to be 'carbon negative' by 2030, and captures the collision between the AI boom and Big Tech's climate promises." category: "Companies" category_url: https://boursel.com/category/companies author: "Kenji Nakamura" published: 2026-07-10T01:37:14.000Z updated: 2026-07-10T01:37:14.000Z canonical: https://boursel.com/article/microsofts-carbon-emissions-jumped-25-percent-as-ai-data-centers-boomed tags: ["microsoft", "emissions", "ai", "data-centers", "climate"] --- # Microsoft's Carbon Emissions Jumped 25% as AI Data Centers Boomed Microsoft's carbon emissions rose about 25% last year, its sustainability report shows, as a frenzy of AI data-center building drove up the concrete, steel and power behind its cloud. The jump pushes the company further from its 2020 pledge to be 'carbon negative' by 2030, and captures the collision between the AI boom and Big Tech's climate promises. The environmental bill for the AI boom is coming due, and Microsoft's is a striking example. The company's carbon emissions climbed roughly 25% last year, [Fortune reported](https://fortune.com/2026/07/09/microsoft-carbon-emissions-2025-data-centers/), citing Microsoft's own sustainability disclosures, as the company raced to build the data centers that power its AI ambitions. It is a sharp move in the wrong direction for a firm that made one of the tech industry's boldest climate promises. ## Why data centers drive emissions The cause is not mysterious: computing infrastructure. Training and running large AI models requires enormous data centers, and both building and operating them are carbon-intensive. Construction devours concrete and steel, materials whose manufacture releases large amounts of carbon dioxide. Operation devours electricity, often more than a small city's worth per large facility. That distinction matters for how the pollution is counted. Companies split emissions into three "scopes": Scope 1 is direct emissions from their own operations, Scope 2 is the electricity they buy, and Scope 3 is everything upstream in their supply chain, including the carbon baked into construction materials and hardware. For Microsoft, the great bulk of its footprint sits in Scope 3, which is exactly why a data-center building spree shows up as higher emissions even when the finished facilities run partly on clean power. ## The pledge under strain The increase is awkward because of what Microsoft promised. In 2020, the company committed to becoming "carbon negative" by 2030, removing more carbon from the atmosphere than it emits, and even to erasing, by 2050, all the emissions it has produced since it was founded in 1975. Those were among the most ambitious targets in corporate America. Emissions rising a quarter in a single year, rather than falling toward zero, makes the 2030 goal look increasingly difficult. Microsoft has not abandoned the effort. It has signed large deals for renewable power and, notably, an agreement to help restart a nuclear reactor at Three Mile Island in Pennsylvania to feed its data centers with carbon-free electricity, and it continues to buy carbon-removal credits. But even those steps are struggling to keep pace with how fast its computing demand, and its construction, are growing. ## A whole-industry problem Microsoft is not an outlier; it is the rule. The other giants racing to build AI are reporting the same thing. Google's emissions also jumped sharply as its data-center electricity use surged, and Amazon and Meta have reported rising footprints too, [TechCrunch reported](https://techcrunch.com/2026/07/02/a-warning-sign-about-ais-real-cost-courtesy-of-google-and-amazon/). Each had made climate pledges that now look harder to hit. The common thread is structural: demand for AI computing is outrunning the available supply of clean electricity, and no amount of corporate goodwill closes that gap on its own. ## Why it matters Beyond the optics, this is a real economic and policy story. The AI build-out is colliding with power grids, with the emissions embedded in global manufacturing, and with the still-immature market for carbon removal, all at once. For investors, it raises a question that will only grow louder: whether the companies spending hundreds of billions on AI can reconcile that spending with the climate commitments they made in a different era, and what it costs, financially and reputationally, if they cannot. Microsoft's 25% jump is one data point, but it points to a tension the whole industry will have to confront. This article is informational and not investment advice. ## Sources - [Microsoft's emissions surged 25% in 2025 during data center boom](https://fortune.com/2026/07/09/microsoft-carbon-emissions-2025-data-centers/) - [A warning sign about AI's real cost, courtesy of Google and Amazon](https://techcrunch.com/2026/07/02/a-warning-sign-about-ais-real-cost-courtesy-of-google-and-amazon/)