By
Stuart Kerr, Technology Correspondent,
LiveAIWire

The carbon footprint of AI systems
alone could be equivalent to that of New York City in 2025, with emissions
between 32.6 and 79.7 million tonnes of CO2, while the water footprint could
reach 312.5 to 764.6 billion litres, according to research published in the
journal Nature
Sustainability in November 2025. Data centre electricity
consumption reached an estimated 415 terawatt-hours in 2024, a figure the IEA
projects will rise to 945 terawatt-hours by 2030, when data centres,
cryptocurrency, and AI could collectively account for approximately 4 percent
of annual global energy usage, roughly equal to Japan’s entire electricity
consumption. These are the numbers that define the environmental cost side of
the AI equation, and they are large enough to require serious engagement
rather than dismissal.

But they are also only half the
equation, and the half that dominates public coverage is not the half that
determines whether AI’s net effect on the environment is beneficial or
harmful. The other half is what AI enables: renewable energy optimisation, grid
management, predictive maintenance for clean infrastructure, climate
modelling, and emissions reduction applications that the IEA estimated could
deliver environmental benefits substantially exceeding AI’s own footprint if
deployed at scale. The environmental story of AI in 2026 is a genuine
paradox, not resolvable by citing only the cost or only the benefit, and
understanding both sides with the specificity the numbers deserve is the
precondition for making good decisions about it at every level from
individual tool choice to national energy policy.

The Cost
Side: What the Data Actually Shows

The Nature
Sustainability research clarifies an important distinction that most coverage
misses: inference, not training, drives 80 to 90 percent of AI’s ongoing
energy use. The energy required to train a large AI model is significant but
one-time. The energy required to run billions of daily user interactions with
that model accumulates continuously and grows with adoption rather than
staying fixed. This means AI’s energy footprint is directly coupled to its
success: the more useful AI becomes and the more it is used, the larger its
energy consumption grows, independent of improvements in training
efficiency.

The United
Nations University analysis published in June 2026 adds the land
dimension to the picture. Global data centre electricity use, estimated at
448 terawatt-hours in 2025, carries an associated water footprint projected
at 9.3 trillion litres by 2030 and a land footprint of over 14,500 square
kilometres for the facilities required. That land figure is roughly the size
of Connecticut, devoted to the physical infrastructure required to run AI
systems at their projected 2030 scale. The environmental disclosure gap
compounds the measurement challenge: insufficient transparency from data
centre operators makes precise attribution of environmental impact to AI
versus non-AI workloads difficult, which means current estimates carry large
uncertainty ranges and could be wrong in either
direction.

The Benefit Side: What the Same Technology
Enables

Google reported in 2025 that it had reduced the
median energy consumption per Gemini prompt by a factor of 33 and the
associated carbon footprint by a factor of 44 within a single year of
deployment optimisation. That efficiency improvement rate, if sustained,
changes the long-term energy trajectory substantially. More significantly for
the net environmental calculation, the applications that AI enables in energy
systems, as covered in detail in how
AI is accelerating renewable energy deployment, are already
delivering quantified benefits: the IEA identified up to 175 gigawatts of
additional transmission capacity unlockable through AI grid optimisation, and
up to $110 billion in annual power plant cost savings by
2035.

The asymmetry between the environmental cost of AI
and the environmental benefit it enables is not straightforwardly positive or
negative. It depends on whether AI’s energy growth is powered by clean or
fossil sources, whether the renewable and efficiency applications are
actually deployed at the scale needed to offset the demand growth, and
whether the efficiency improvements in inference continue at their current
rate or plateau. All three are uncertain in ways that make confident claims
about AI being net good or net bad for the environment
premature.

What Needs to Change to Resolve the Paradox

The
resolution of the AI environmental paradox depends on three converging
decisions being made right now. First, data centre siting and energy
procurement: the carbon intensity of AI workloads varies dramatically
depending on where the data centres are located and what energy sources they
use. Microsoft’s 2025 environmental report noted the launch of a new data
centre design that uses zero water for cooling, a significant improvement for
water-stressed regions. Second, disclosure requirements: the insufficiency of
current environmental reporting from AI companies means that the market
cannot currently price environmental cost accurately, which removes one of
the mechanisms that would otherwise drive efficiency investment. Third, regulatory
alignment: the Paris Agreement targets a 53 percent reduction in data centre
emissions by 2030, a target that requires coordinated policy rather than
voluntary efficiency efforts.

The AI environmental story is
not a stable situation. It is a race between the growth of AI energy demand
and the decarbonisation of the energy systems that supply it. Understanding
the scale of the
investment flowing into AI infrastructure is the starting point for
understanding why the demand side of that race is not going to slow down
voluntarily. The supply side, renewable energy and grid decarbonisation,
needs to keep pace with deployment that the private sector is funding at a
scale that public investment in clean energy has not historically matched.
Whether it does will be determined more by energy and climate policy over the
next five years than by anything the AI industry decides independently. The
expansion of ambient AI into everyday environments represents a
further growth vector for AI energy demand that most environmental analyses
have not yet fully incorporated.

The Disclosure Gap That
Prevents Honest Accounting

The most important practical
problem in the AI environmental debate is not the scale of the impact but the
quality of the data. The Nature Sustainability research was explicit that
insufficient environmental disclosure from data centre operators makes
precise attribution of environmental impact to AI versus non-AI workloads
impossible with current information. Companies report aggregate data centre
energy and water use; they do not, in most cases, separately report what
portion of that consumption is driven by AI workloads versus general
computing. This means the uncertainty ranges on AI environmental footprint
estimates are wide enough to span genuinely different policy conclusions: at
the lower end of the range, AI’s footprint is significant but manageable with
existing decarbonisation trajectories; at the upper end, it threatens to
undermine Paris Agreement targets for the ICT sector.

The
ISO/IEC TR 20226:2025 standard published in 2025 provides an overview of
environmental sustainability metrics for AI systems across their full
lifecycle, including water footprint as a key metric. This is the standard
infrastructure that will allow more precise accounting, but standards cannot
substitute for the political will to mandate the disclosure they enable. The
gap between what could be measured and what is being disclosed is itself a
policy failure, and closing it is the prerequisite for any honest resolution
of whether AI’s net environmental effect is what its proponents claim or its
critics fear.

About the Author

Stuart
Kerr is Technology Correspondent at LiveAIWire, covering artificial
intelligence, cybersecurity, and the social impact of emerging technology. He
publishes daily at LiveAIWire.com.