Industry Impact2 min read

Microsoft's Carbon Emissions Up 25% as AI Demands More Power

July 10, 2026Synthesized from 2 sources: The Guardian, Engadget

Microsoft's own sustainability report shows its carbon output jumped 25% in one year, driven almost entirely by the energy needed to run AI systems, putting its 2030 carbon-negative pledge in serious doubt.

Microsoft published its 2026 sustainability report this week. The headline finding is a 25% rise in total carbon emissions in fiscal year 2025. The company emitted roughly 20 million metric tons of carbon dioxide equivalent, up from 16 million the year before. The primary cause is straightforward: it is building more data centers, faster, to run AI services, and those buildings consume enormous amounts of electricity.

What makes this more than a Microsoft story is that every major technology company is reporting the same thing at the same time. Amazon's emissions rose 16% last year. Google's rose 18%. Meta had previously reported a 64% jump. These companies together control roughly two-thirds of the data center power capacity among the top 15 players tracked by analysts at Jefferies. When their emissions go up, the sector's footprint goes up with them.

The fundamental problem is a timing mismatch. Companies can build new data centers in two to three years. Building new clean power plants, upgrading power grids, and developing new energy storage takes much longer. So when demand spikes faster than clean energy can scale, the gap is filled with natural gas and other fossil fuels. Microsoft itself just signed a 20-year agreement with Chevron to power a single new data center complex in West Texas using a gas plant that will generate up to 2.67 gigawatts, enough electricity to power around two million homes.

Microsoft pledged in 2020 to become carbon negative by 2030, meaning it would remove more carbon from the atmosphere than it produces. With four years left, its emissions are substantially higher than when it made that pledge, and it plans to keep spending. Capital expenditures this year are set at $190 billion, a 61% increase over last year, largely for new data centers.

There are two things worth understanding about why the reported emissions number jumped as sharply as it did. First, the genuine increase from new building and energy use. Second, a change in accounting: Microsoft stopped buying a type of short-term renewable energy certificate that had previously reduced its reported emissions figure. The company argues this change is actually better for the environment long-term, because it pushes toward building real clean energy capacity rather than just buying paper credits. That logic is reasonable, but the immediate effect is a higher number in the report.

For business operators outside the tech sector, the practical takeaway is about energy costs and supply. When the largest companies in the world compete for electricity at this scale, they put pressure on local power grids, push up energy prices, and in some regions create real supply constraints. Cornell researchers estimate that if AI growth continues at its current rate, data centers in the US alone could require the equivalent of adding five to ten million cars' worth of carbon emissions annually by 2030. That is not a distant problem for someone else. If your operations depend on stable, affordable energy, or if your suppliers do, this trajectory is worth tracking.

The water question is also real. Data centers use water to cool their servers. Microsoft did report that globally it replenished more water than it withdrew last year, which is a genuine step forward. But total water use is still rising alongside AI growth, and many data centers are being built in water-scarce regions.

Microsoft is not hiding from any of this. Its report uses direct language: sustainability solutions are not scaling fast enough to meet demand. That is an honest admission. The question is whether the gap closes fast enough to matter by 2030, or whether the 2030 target quietly becomes a 2035 target.

Stay informed

Get AI intelligence like this delivered to your inbox.


You May Also Find Valuable