By Daily Touch Insights Editorial Team
Editorial Team
View Journalist Profile
TECHNOLOGY & ENVIRONMENT — The explosive expansion of artificial intelligence is creating a new environmental challenge for the world's largest technology companies: how to build enough data centres to support AI without sending their carbon emissions sharply higher.
Amazon, Microsoft, Google and Meta are investing enormous sums in new data-centre infrastructure as demand for AI services, cloud computing and digital storage accelerates. But the electricity and construction required to support that expansion could make it increasingly difficult for technology companies to meet their climate commitments.
Recent analysis of 60 planned data centres found that the facilities could produce about 101.5 million tonnes of carbon dioxide annually, equivalent to roughly 7% of U.S. power-sector emissions in 2025. 0
AI Is Driving an Extraordinary Construction Boom
The AI industry needs far more computing power than traditional internet services.
Training advanced models and operating millions of AI queries require huge quantities of specialised chips, servers and electricity.
That is pushing technology companies to build increasingly large data centres capable of housing thousands of AI processors.
The scale of the investment is enormous, with the world's largest technology companies expected to spend hundreds of billions of dollars on AI infrastructure in 2026. 1
Big Tech's Emissions Are Already Rising
The environmental impact is no longer simply a future prediction.
Microsoft, Amazon and Google collectively produced about 119 million metric tonnes of carbon-dioxide equivalent in the financial year ending March 2026, nearly one-fifth more than the previous year. 2
The increase demonstrates the difficulty of reducing emissions while simultaneously expanding the infrastructure required for AI.
Microsoft Provides a Clear Example
Microsoft reported that its carbon emissions increased by 25% in 2025, reaching approximately 20 million metric tonnes of carbon-dioxide equivalent.
The company attributed the increase largely to new data-centre construction and changes involving renewable-energy credits. 3
This creates a direct conflict between Microsoft's rapidly expanding AI business and its long-term environmental ambitions.
More Data Centres Mean More Electricity
Data centres operate around the clock.
Servers must remain active, cooling systems must operate continuously and networking equipment must handle enormous quantities of information.
AI workloads make the problem even larger because advanced processors can consume significant amounts of electricity while performing intensive calculations.
If electricity demand grows faster than clean-energy capacity, utilities may have to rely more heavily on natural gas and other fossil fuels.
Natural Gas Could Become the Unexpected Winner
The AI boom is already creating pressure on electricity grids in several regions.
Utilities facing rapidly increasing demand may turn to natural-gas power plants because they can often be built or expanded more quickly than some renewable and transmission projects.
That creates a paradox: companies may be investing heavily in AI while indirectly increasing demand for fossil-fuel electricity.
Some proposed data centres are even considering dedicated gas generation to secure reliable power while waiting for broader grid infrastructure to catch up. 4
Renewable Energy Does Not Automatically Solve the Problem
Technology companies have committed billions of dollars to renewable energy.
Google and Microsoft, for example, have ambitious targets for reaching net-zero emissions.
But building renewable generation does not necessarily mean that every data centre is operating entirely on clean electricity at every moment.
When wind and solar generation are unavailable, data centres still need electricity from the wider grid or other sources.
The timing between electricity consumption and clean-energy production therefore matters.
Construction Itself Creates Emissions
The environmental footprint of a data centre does not begin when its servers are switched on.
Constructing massive buildings requires steel, concrete, glass, electrical equipment, cooling systems and other materials.
Those materials have their own carbon footprints.
As companies build thousands of new facilities simultaneously, construction-related emissions can become a significant part of the technology industry's overall environmental impact.
The Cloud Is Not Weightless
Consumers often think of cloud computing as something that exists virtually.
In reality, every photo, video, application, AI conversation and business database stored in the cloud ultimately depends on physical infrastructure.
That infrastructure includes buildings, servers, power systems, cooling equipment and networks.
The growth of cloud computing therefore creates a physical environmental footprint that users rarely see.
AI Is Making the Problem Larger
Traditional cloud services already required substantial computing infrastructure.
Generative AI adds another layer of demand.
Large language models must be trained using powerful computing clusters, while serving those models to millions of users requires continuous inference capacity.
As AI becomes integrated into search engines, office software, coding tools, advertising systems and consumer applications, electricity demand could continue increasing.
Big Tech's Climate Targets Are Under Pressure
The industry's climate ambitions have not disappeared.
Google, Microsoft and Amazon continue to maintain long-term emissions targets.
But the rapid expansion of AI infrastructure is making those targets harder to achieve.
The fundamental problem is simple: companies are attempting to reduce the environmental impact of their operations while simultaneously building much more infrastructure.
There Is a Difference Between Accounting and Physical Emissions
Another challenge involves how companies calculate their carbon footprints.
Corporate climate reporting can include renewable-energy certificates and other mechanisms designed to account for clean-energy purchases.
Critics argue that these accounting methods can make a company's reported emissions appear lower than the physical emissions associated with the electricity actually consumed by its facilities.
That debate is becoming increasingly important as data-centre electricity demand grows.
AI Could Also Improve Efficiency
The environmental story is not entirely negative.
AI can potentially help electricity grids forecast demand, optimise industrial processes, improve renewable-energy generation and reduce waste.
More efficient AI chips and better data-centre cooling systems could also reduce the amount of electricity required for individual workloads.
The problem is that efficiency improvements can be overwhelmed if total demand grows even faster.
The Rebound Effect Is a Serious Risk
If AI becomes cheaper and more efficient, people and businesses may simply use more of it.
A model that requires half as much electricity per query may not reduce total electricity consumption if the number of queries increases fivefold.
This means technological efficiency alone cannot guarantee lower emissions.
Data Centres Are Becoming an Energy Policy Issue
The rapid growth of AI infrastructure is forcing governments and utilities to reconsider how electricity grids are planned.
Large technology companies can require enough power to influence local energy markets and infrastructure decisions.
That raises difficult questions about who should pay for new power plants, transmission lines and grid upgrades required to support private technology projects.
Communities Are Also Feeling the Pressure
Data centres require more than electricity.
Many facilities also require significant amounts of water for cooling, while their construction can affect land use, local infrastructure and surrounding communities.
As the number of facilities increases, residents and policymakers are increasingly questioning whether the economic benefits justify the environmental and infrastructure costs.
The AI Industry Cannot Outsource Its Carbon Footprint
One of the biggest problems is that companies using cloud AI services can effectively shift part of their environmental footprint to the technology companies operating the infrastructure.
A business may reduce its own physical computing infrastructure while increasing its dependence on cloud providers.
The emissions do not disappear. They simply move elsewhere in the supply chain. 5
The Numbers Could Become Much Larger
The current data-centre boom is still developing.
Hundreds of facilities are being planned or constructed to meet expected AI demand.
If electricity grids fail to decarbonise quickly enough, the emissions associated with this infrastructure could remain substantial for years.
The outcome will depend heavily on where new data centres are built and what sources of electricity power them.
What Big Tech Needs to Do
Technology companies cannot rely exclusively on promises to purchase renewable energy.
They will need to increase the amount of genuinely clean electricity available to the grids where their data centres operate.
They can also invest in more efficient chips, advanced cooling systems, battery storage, low-carbon construction materials and better workload management.
Most importantly, their climate commitments need to account for the physical expansion required by AI.
Governments Face a Difficult Choice
Governments want the economic benefits associated with AI investment.
New data centres can bring construction activity, technology investment and tax revenue while strengthening a country's digital infrastructure.
But governments also have climate targets and limited electricity resources.
The challenge will be allowing AI infrastructure to expand without allowing private technology demand to undermine broader decarbonisation efforts.
Our Perspective
The biggest mistake would be to assume that AI is automatically environmentally friendly simply because it is digital.
AI may help reduce emissions in some industries, but the infrastructure powering it is physical, energy-intensive and increasingly enormous.
The current construction boom shows that the environmental cost of AI is becoming impossible to ignore.
The real test for Big Tech is no longer whether it can build enough data centres to win the AI race. It is whether it can build them without turning the electricity and climate systems into the hidden cost of that victory.
Conclusion
The rapid expansion of AI data centres is putting increasing pressure on Big Tech's climate ambitions.
Microsoft, Amazon and Google have already reported significant increases in emissions as their infrastructure expands, while planned U.S. data centres could add a substantial amount of annual carbon emissions if powered by carbon-intensive electricity. 6
Renewable-energy investment and efficiency improvements can reduce the impact, but they may not be enough if electricity demand continues growing at the current pace.
The AI revolution is ultimately an energy revolution as well. Whether it becomes compatible with global climate goals will depend not only on smarter algorithms, but on how the world chooses to power the machines running them.
