Big Tech’s Carbon Emissions Surge as Datacentre Construction Accelerates
The collective carbon emissions of Microsoft, Amazon, and Google increased by nearly a fifth over the past year, driven primarily by a massive expansion of datacentre infrastructure to support artificial intelligence. The three companies emitted 119 million metric tonnes of carbon dioxide equivalent (mTCO₂e) in the financial year ending March 2026 — roughly a third of France’s total emissions. The previous year, they emitted approximately 101m mTCO₂e, comparable to Czechia’s 2024 emissions.
A Reversal of Years of Progress
The figures, documented in annual sustainability reports released over recent weeks, mark a stark reversal for companies that had spent years publicly committed to reducing their carbon footprints. Microsoft’s emissions had appeared to flatline at 16m mTCO₂e across 2023 and 2024. This year, Microsoft reported a 25% increase to 20m mTCO₂e, which the company attributed “primarily” to “the expansion of our datacentre infrastructure.”
Google reported an 18% emissions increase, citing “increases in supply chain activities that supported the rapid expansion of our business.” Amazon reported a 16% increase overall and a 20% increase in supply chain emissions, which included datacentre building and construction. Despite these increases, Amazon framed its report as “making progress” toward its goal of net zero emissions by 2040. All three companies maintain their net zero commitments: Google and Microsoft by 2030, Amazon by 2040.
Google also noted that its AI systems have produced solutions that helped reduce emissions elsewhere by 41m tonnes of CO₂ last year — a claim that underscores the companies’ broader argument that AI can be deployed as a tool for environmental good.
The AI Infrastructure Spending Boom
The emissions surge is tightly linked to what has become the largest infrastructure spending drive in the technology sector’s history. The world’s biggest tech companies are on track to spend $765bn (£570bn) this year, the majority of it directed toward building AI datacentres in locations ranging from Norway to North Tyneside.
JLL, a US property consultancy, projects that approximately 1,200 datacentres will be built globally between now and 2030, with demand “overwhelmingly driven by AI.” The Uptime Institute, which rates and inspects datacentres, estimates that big datacentre projects announced last year alone would consume 1.3% of the world’s electricity usage — nearly doubling current datacentre power demand. The majority of that new power demand will come from US projects.
Carbon Credit Supply Under Strain
Shaolei Ren, a professor of electrical engineering at the University of California, Riverside, said the emissions increases are “strongly correlated” with the companies’ AI investments. He also flagged a secondary concern: Microsoft’s sustainability report suggested fewer carbon credits were available on global markets to offset emissions.
“While companies are actively investing in or purchasing carbon credits, the figure suggests a possible lack of credit supply in the carbon market to meet the technology companies’ needs,” Ren said. “Everyone is talking about the lack of physical goods and infrastructure like power, but there may also be a lack of virtual goods — carbon credits.”
Outsourcing Emissions to the Cloud
Cecilia Rikap, an economics professor at University College London, offered a more critical assessment of the companies’ sustainability claims. “Claims by Microsoft, Amazon and Google about their clouds being ecologically friendly and sustainable are a marketing strategy,” she said. “Governments should remember these expanding carbon footprints when the very same companies offer addressing the ecological crisis with AI solutions.”
Rikap also pointed to a structural issue in how corporate emissions are accounted for. As more companies migrate operations to cloud platforms and use them to train and deploy AI models, those companies effectively outsource their digital carbon footprints to the cloud providers. “Basically, shifting to the cloud helps other corporations obscure their environmental footprint,” she said.
Microsoft, Google, and Amazon were contacted for comment.
Context: The Scale of the Problem
The numbers place the three companies’ combined emissions in a context typically reserved for nation-states. At 119m mTCO₂e, their annual output exceeds the total emissions of many individual countries and rivals the output of mid-sized European economies. The year-over-year jump of roughly 18m mTCO₂e — from 101m to 119m — alone approaches the annual emissions of several smaller developed nations.
What distinguishes this emissions growth from earlier periods is its direct linkage to a single technological trend. The cloud computing boom of the 2010s expanded tech companies’ footprints, but the current surge is specifically tied to the construction of purpose-built AI infrastructure — facilities designed to house the specialized processors required to train and operate large language models and other AI systems. These facilities carry energy and construction demands that exceed those of conventional datacentres.
With $765bn in planned spending this year and JLL projecting 1,200 new datacentres by 2030, the trajectory suggests emissions will continue rising unless the companies can secure sufficient clean energy and carbon offsets — or unless they revise their net zero timelines. For now, all three maintain their targets, even as the gap between current emissions trajectories and those commitments widens.