AI and Climate Change: Promise, Peril, and the Path Forward

As a sustainability professional interested in artificial intelligence (AI), I feel like we now live in an era of two accelerating forces. On one hand, AI is reshaping industries at breathtaking speed, driving efficiencies and breakthroughs once unimaginable. On the other hand, climate change is intensifying, exposing the vulnerabilities of our food, water, and energy systems, as well as testing the resilience of our communities. The intersection of these two megatrends is not coincidental — AI is both part of the climate problem and potentially one of its most effective solutions. Understanding this paradox is critical if we want to harness digital power for planetary resilience.

The Climate Cost of Artificial Intelligence

The narrative often starts with AI’s potential, how it can cut millions of dollars and expedite the product development cycle. However, what about its footprint? Data centres already consume staggering amounts of energy — about 415 terawatt-hours (TWh) in 2024, roughly 1.5% of global electricity demand. Projections suggest this could more than double to 945 TWh by 2030, equivalent to the annual consumption of Japan. Water is another hidden cost: a single large model training run can consume hundreds of thousands of litres for cooling. Researchers estimate that even casual use of generative AI (10–50 queries) consumes about 500 mL of water, invisible but real.

Global data centre electricity consumption, by equipment, Base Case, 2020 - 2030 (Source: IEA 2025).

This is not trivial when juxtaposed with global net-zero goals. In regions where power grids are still dominated by coal or gas, data centre demand risks locking in carbon-intensive infrastructure. Some U.S. utilities have already delayed coal plant retirements to meet surging AI demand. If left unchecked, AI could exacerbate the very problem it is expected to help solve.

The Climate Dividend: AI as a Catalyst for Decarbonization

The same technology that strains our grids may also save them. The London School of Economics recently estimated that AI applications across energy, transport, and food systems could potentially cut 3.2 to 5.4 billion tonnes of CO₂-equivalent annually by 2035 — a figure comparable to eliminating the combined yearly emissions of India and Japan.

Concrete use cases are emerging fast:

• Smart grids that use AI to forecast demand and integrate intermittent renewables like solar and wind.

• Precision agriculture, where AI-powered sensors and drones optimize use of water and fertilizer, reducing both emissions and waste, and minimizing pollution of water bodies due to run offs.

• Logistics optimization, where machine learning reduces empty freight runs and optimize the loading volume, cutting transport emissions.

• Industrial efficiency, with AI accelerating discovery of low-carbon materials, batteries, and catalysts that would take years to identify in traditional labs.

• Environmental biotechnology, where AI can be used to find innovative microbes and microbiomes that would help to degrade toxic materials in the environment.

  
  

In each of these areas, AI acts as a force multiplier: it accelerates pattern recognition, simulates scenarios, and optimizes decisions at scales humans cannot manage alone.

New Climate Intelligence: Satellites, Simulations, and Anticipation

One of AI’s most transformative applications is in climate intelligence. Satellite constellations now produce petabytes of Earth observation data daily. Without AI, this deluge is noise; with AI, it becomes actionable. Startups and research groups are mapping methane leaks, illegal deforestation, and even real-time industrial emissions — building transparency where previously there was opacity.

At the frontier, Nvidia’s “Climate-in-a-Bottle” promises 5 km resolution climate models, compared to the coarse 25–50 km grids of traditional systems. That level of detail could inform city-level adaptation planning and real-time disaster anticipation. Meanwhile, AI-driven early warning systems are moving from forecasts to anticipatory action: identifying vulnerable communities and pre-positioning resources before floods, droughts, cyclones strike, or other natural disasters. While the technology is still at budding stage, it is definitely needed with the increasing frequency of extreme weather events. In fact, this is not just incremental progress. It is a qualitative shift in how we can understand, predict, and manage systemic climate risks.

Digital Twin of Earth (Source: WSJ, by courtesy of Nvidia)

The Equity Challenge: Who Benefits? Who Bears the Cost?

These advances, however, carry risks of inequity. AI innovation is concentrated in a handful of tech firms and wealthy nations, yet its environmental costs are global. Data centers are often sited in regions with cheaper power and water, raising questions of climate justice: who bears the local burden of water withdrawals, emissions, and land use to train models that may primarily benefit high-income economies?

Moreover, the climate benefits of AI are not automatic. Without supportive governance, efficiency gains can lead to rebound effects — for example, cheaper logistics that increase overall freight traffic, or more efficient cooling systems that prompt more widespread adoption of energy-hungry air conditioning. In other words, AI can reduce emissions in theory while increasing them in practice.

Regulatory frameworks lag far behind the pace of AI development. The U.S. Artificial Intelligence Environmental Impacts Act of 2024 was a first attempt to mandate reporting of energy and water use for large models, but global coverage remains patchy. The EU’s Destination Earth program is perhaps more ambitious: building a “digital twin” of the planet to guide climate and resource management. Yet the gap between visionary projects and widespread accountability is wide.

Corporate initiatives like Microsoft’s AI for Earth ($50 million, nearly 1,000 projects funded across 40 countries) show potential but remain voluntary and piecemeal. To fully align AI with climate goals, stronger disclosure, incentives, and shared governance will be necessary.

Toward Climate-Conscious AI

The paradox of AI and climate change is not whether it will be part of the solution — it already is. The real question is whether it will become part of the problem in equal or greater measure. To avoid that, three priorities emerge:

1. Transparency and Reporting

Data on energy, water, and emissions use must be disclosed for all large-scale AI models. Without it, accountability is impossible.

2. Green Infrastructure

Rapid expansion of renewable energy and low-water cooling for data centers should be treated as critical infrastructure, not optional add-ons.

3. Directed Innovation

Incentives should steer AI research toward high-impact climate applications: renewable integration, biodiversity monitoring, climate finance modeling, and adaptation planning.

If we pursue AI without climate consciousness, we risk reinforcing existing vulnerabilities. If we pursue climate action without AI, we risk leaving powerful tools unused. But if we integrate the two — responsibly, equitably, and at scale — AI could become one of the most important accelerants in humanity’s fight for resilience.

The story of AI and climate is not just about vibe coding or carbon emission. It is about governance, values, and choices. The future is being written in both silicon and atmosphere. Just as in education or medical field, whether AI becomes a climate burden or a climate ally will depend on whether we make those choices deliberately, and soon.

[First published on Substack "Ginci Insights" on August 20, 2025: https://gincinno.substack.com/p/ai-and-climate-change-promise-peril?r=2cxt8s]