December 2, 2025. — Susan Greene

• Mail
• Facebook
• Threads
• LinkedIn
• Bluesky
• Print

The 2025 Economist article, “How AI Could Help the Climate,” highlights the promise of artificial intelligence to drive climate mitigation and adaptation. Yet emerging evidence shows that AI’s infrastructure — particularly data centers — can consume vast amounts of energy, draw heavily on water resources, and strain local ecosystems. How we design and deploy these systems today will determine whether AI becomes a net benefit or a new environmental challenge.

The Climate Promise of AI

AI’s strengths — data processing, prediction, optimization, pattern detection — align with many of the most urgent needs of a warming world. Experts and organizations note that AI can:

• Help manage and stabilize power grids so that more renewable energy (wind, solar) can be rapidly and reliably integrated, compensating for variability and reducing fossil-fuel reliance.
• Optimize logistics and supply chains to cut fuel use, emissions, and waste in transport, manufacturing, and food systems.
• Improve agricultural efficiency and support sustainable farming by using predictive models for weather, soil health, resource use, and yield — potentially cutting emissions from food systems while conserving resources.
• Enable better climate-resilience planning: from forecasting extreme weather and disasters, to supporting adaptation efforts through data analysis, resource allocation, and monitoring.

These are not pipe dreams. As The Economist argues, AI is a tool that — under the right policies and deployment — can help speed up the energy transition, reduce waste, and enhance resilience.

But AI’s Infrastructure is Thirsty: Water and Energy Under Pressure

The flip side is increasingly hard data that shows AI’s growth driving resource pressures — especially on water and energy.

• A 2025 report by Bloomberg highlights a surge: more than 160 new AI-specialized data centers have appeared across the U.S. in recent years — many in regions already under water stress.
• Data-center cooling — essential because of the massive heat generated by AI servers — consumes substantial amounts of fresh water. In some data-center designs, each kilowatt-hour of energy use translates into roughly 2 to 2.4 gallons of water consumed for cooling.
• A comprehensive recent review points out that the water burden isn’t just “a bit more” — it can be enormous. Under certain growth trajectories, data-center water withdrawals could stress drinking-water supplies, agricultural water, and local ecosystems in drought-prone regions.
• In one modelling study, researchers projected that by 2030 the AI-driven data-center boom could withdraw between 731 and 1,125 million cubic meters of water per year — roughly equivalent to the household water use of 6–10 million Americans. At the same time, electricity-related emissions from these centers could add 24–44 million metric tons of CO₂ annually (equivalent to millions of cars on the road).

In short: AI’s rise comes with growing demands on water, energy, and the broader environmental systems that supply them.

Environmental Justice, Local Impacts, and Unequal Burdens

Because data centers are not evenly distributed — they tend to cluster where power is cheap, infrastructure exists, and regulations are permissive — the environmental burden can be highly uneven. Recent reporting warns that many new facilities are being built in water-scarce regions, potentially drawing down groundwater, competing with agriculture or local drinking-water, and worsening drought risk.

Local water utilities, often already under pressure, may lack the capacity to handle huge withdrawals or wastewater. One recent policy analysis calls for regional planning: water-system managers, economic-development agencies, and local governments need to coordinate with tech developers to ensure sustainable siting and water sourcing — not ad-hoc, reactive decisions.

These are equity issues: in many cases, the communities hosting data centers may get limited economic benefit, no jobs, but face outsized environmental risks. Without safeguards, AI expansion could worsen existing disparities in water access, pollution burden, and climate resilience.

Steering AI Toward Climate Benefit — What Needs to Be Done

The evidence suggests that whether AI helps or harms the climate depends less on “AI itself” and more on how we build and manage the supporting infrastructure.

A Conditional Verdict — AI as Opportunity and Responsibility

In fact, the stakes are becoming immediate. A recent Reuters‑reported dispute over a proposed AI data‑center project in rural Pennsylvania shows how quickly these issues can erupt into community backlash. What began as a development plan — part of a broader push for more AI infrastructure — met with strong resistance from local residents concerned about lost farmland, rising utility costs, and environmental impacts.

That pushback underscores a key point: the environmental and social consequences of AI infrastructure are not abstract — they unfold in real communities, often rural or water‑stressed, and can ignite conflict when planning occurs without local input, transparency, or safeguards.

As AI continues to grow, the question isn’t whether data centers will expand — it’s whether they will expand responsibly: guided by clean energy sourcing, water‑smart design, transparent environmental reporting, and equitable community engagement. If not, the same technology with climate‑benefit potential may instead strain ecosystems, disrupt local land use, and trigger resistance — undermining both climate and social goals.

If done right, AI could become a powerful ally in climate mitigation and adaptation. If done carelessly — without attention to water, energy, equity, and long-term impacts — it could deepen environmental stress and inequality.