AI and climate: Paths to sustainable progress

AI and climate are more closely linked than it seems at first glance. Artificial intelligence can help to reduce emissions, use resources more efficiently and better understand climate risks — but at the same time causes energy consumption, CO₂ emissions and water consumption itself. This article shows how AI and climate can be combined without ignoring ecological costs.

‍

‍

‍

AI and climate: Why the issue is now decisive

‍

Climate change is one of the biggest challenges of our time — and at the same time a driver of technological innovation. AI is considered an important lever for reducing emissions, accelerating energy transitions and better modelling complex climate systems. Studies estimate that AI could help save around 3.2—5.4 gigatons of CO₂ equivalents worldwide annually by 2035, particularly in the energy, mobility and industry sectors.

‍

At the same time, resource consumption is exploding in the background: The International Energy Agency expects that the electricity consumption of data centers could approximately double by 2030 — driven by AI workloads. Analyses for Greenpeace assume that AI data centers in particular could increase their power consumption from around 50 TWh in 2023 to around 550 TWh in 2030.

‍

AI and climate are therefore both an opportunity and a risk: Without clear guidelines, AI itself can become a climate problem, but with the right applications and a sustainable infrastructure, it can accelerate transformation.

‍

‍

‍

Where AI is already helping with climate protection

‍

‍

Environmental monitoring and early warning systems

‍

A central intersection of AI and climate is the evaluation of huge environmental data sets. Every day, satellites, sensors and drone networks provide terabytes of information about ice rinks, forests, oceans and cities. AI models recognize patterns that people can no longer understand.

‍

Examples: systems that detect forest fires in satellite images at very early stages, models for monitoring glacier melt, or algorithms that identify plastic waste in oceans. The EU project “Destination Earth” is also working on a digital copy of the planet in order to be able to simulate climate scenarios and prepare political decisions based on data.

‍

This becomes relevant for companies at the latest when supply chains are threatened by extreme weather, droughts or floods — this is where AI can help identify and manage risks at an early stage.

‍

‍

Energy & power grids

‍

In the energy system, AI ensures that renewable energy can be better integrated into the grid. Models predict wind and solar generation, balance demand and supply, and help to optimally control storage systems.

‍

examples:

• Network operators use AI to smooth out load peaks and identify bottlenecks at an early stage.
• Public utilities rely on AI-based forecasts to manage heating networks more efficiently.
• Building management systems dynamically regulate heating, cooling and lighting according to use and outside temperature.

‍

In this way, AI and climate are concretely linked — with measurable effects on the CO₂ balance and energy costs.

‍

‍

Agriculture, resource and circular economy

‍

There are also strong interfaces between AI and climate in agriculture and the circular economy. Precision agriculture uses satellite and sensor data to use water, fertilisers and plant protection products in a targeted and economical way. This reduces emissions, protects soils and increases yields.

‍

In the waste sector, AI systems analyse images of sorting plants, identify materials and thus increase recycling rates. One example: the startup Greyparrot, which claims to reveal enormous amounts of recyclable material that would otherwise end up in incineration.

‍

Such applications show that AI and climate are not just a “tech issue,” but a tangible issue of efficiency and costs.

‍

‍

AI and climate: The ecological footprint of AI systems

‍

As helpful as AI can be for the climate, it is itself resource-intensive. Power consumption, cooling water, hardware production and disposal have a direct impact on the climate balance.

‍

Studies on the “Energy Footprint of AI” indicate that large model training runs can cause several hundred tons of CO₂ when powered by fossil energy. Recent analyses show: While individual AI queries require relatively little energy, hundreds of millions of requests per day add up to considerable consumption.

‍

For example, Google states that a request to the Gemini AI system requires an average of around 0.24 Wh of electricity and a few tenths of a milliliter of water for cooling — OpenAI mentions similar orders of magnitude for its own models. Extrapolated to global use, this already corresponds to the energy requirements of entire cities.

‍

In addition, there is a growing need for specialized chips, servers and cooling systems. A study commissioned by Greenpeace predicts that AI data centers could consume several hundred billion kWh of electricity per year by 2030 if no countermeasures are taken.

‍

AI and climate are therefore also interrelated because every additional computing load causes ecological costs - especially if the electricity mix is not predominantly renewable.

‍

‍

‍

AI and climate: Green AI and climate-friendly data centers

‍

In order to reconcile AI and climate, the “Green AI” concept is moving into focus. The aim is to achieve the same or better results with less computational effort, less energy and lower emissions.

‍

Important levers:

• More efficient models: UNESCO and UCL show that even small changes to architecture, training and inference energy can reduce the energy consumption of large language models by up to 90% without sacrificing quality.
• Green compute in data centers: More efficiency (PUE values), direct use of renewable energy, waste heat recovery and innovative cooling systems are key measures to make AI infrastructure more climate-friendly.
• Regional location selection: Data centers in regions with a high share of renewable energy and good network infrastructure can significantly improve the CO₂ balance.

‍

Many hyperscalers have now committed themselves to climate neutrality or even “carbon negative” goals and are investing massively in renewable energy and efficiency programs. At the same time, environmental organizations are calling for more transparency on AI-specific emissions — currently these are often only hidden in aggregated sustainability reports.

‍

‍

‍

Sustainable use cases

‍

In order for AI and climate in companies to go together, it comes down to use case on. The decisive factor is whether the benefits achieved by AI (e.g. fewer trips, less waste, less energy) are greater than the additional resource consumption of the AI itself.

‍

Typical “climate-positive” AI scenarios in companies:

• Energy and building management: AI optimizes heating, ventilation and air conditioning systems in office or production buildings and reduces consumption by double-digit percentages.
• Fleet and route optimization: Algorithms plan delivery routes in such a way that mileage, traffic jams and idle times are minimized.
• Supply chain emissions: AI identifies CO₂ hotspots in supply chains, simulates alternatives (other delivery routes, means of transport, suppliers) and supports the selection of climate-friendly options.
• Production optimization: In industry, AI reduces waste, avoids downtimes and controls processes in such a way that energy consumption per unit decreases.
• ESG reporting: AI collects, harmonizes and analyses sustainability data, which improves transparency and governance.

‍

Smaller ESG teams in particular benefit from AI-supported data collection and processing — provided that data protection and governance are properly regulated.

‍

‍

‍

Risks of greenwashing and rebound effects

‍

Where AI and climate are presented as a “dream couple,” the road to greenwashing is not far off. Companies are using “AI for the climate,” while at the same time using energy-intensive models without a clear sustainability strategy. Environmental organizations are therefore warning against too uncritical AI hype.

‍

Two points are particularly critical:

1. Rebound effects: If AI makes processes more efficient, it can result in more being produced or consumed overall — absolute emissions increase despite efficiency. Examples include automated advertising campaigns that drive additional consumption, or ultra-personalized offers that trigger more trips or deliveries.
2. Incomplete accounting: Only direct emissions from the use of AI are often considered, but not emissions from hardware production, disposal or upstream power generation.

‍

For AI and climate to come together credibly, companies should submit their sustainability statements on complete, verifiable data support — and reveal how they minimize the AI footprint.

‍

‍

AI and climate: What companies can do in practice

‍

For the use of AI and climate to work in the same direction, a structured approach is worthwhile. A few practical steps:

1. Prioritize climate-relevant use cases
   Use AI initially where direct emissions or resource savings are expected — for example in energy management, logistics, or process optimization.
2. Capture the CO2 footprint of AI
   Ask your providers specifically for emission data (Scope 2, if applicable Scope 3) for services used. Use existing benchmarks and models to at least roughly estimate additional energy and water consumption.
3. Include green compute criteria in tool selection
   Evaluates AI platforms not only based on functionality, but also on energy efficiency, energy source, transparency reports, and data center location.
4. Encourage efficient use
   Training helps you to formulate precise prompts, bundle multiple requests and avoid unnecessary workload. What sounds trivial can save a lot of energy.
5. Integrate AI into climate strategy
   Explicitly anchor AI in your climate and ESG strategy: What goals does it support, what risks does it entail, and how is its impact measured?

‍

When AI and climate are thought of together in this way, a clear framework is created in which the pressure to innovate and climate protection do not have to be pitted against each other.

‍

‍

‍

AI and climate: Frequently asked questions (FAQ)

‍

‍

Is AI more part of the problem or the solution for the climate?

Both AI can enable significant emissions reductions — studies suggest potentially up to several gigatons of CO₂ equivalent per year by 2035 — but it itself causes considerable energy, water and resource consumption. Whether AI and climate play together positively overall depends on which use cases are implemented and how sustainable the underlying infrastructure is.

‍

How big is the power consumption of AI really?

Specific figures are difficult because many providers only disclose parts of their data. Studies estimate that AI will already be responsible for around 20% of the global power consumption of data centers by the mid-2020s — and the trend is rising. AI data centers are forecasting several hundred TWh of electricity consumption per year by 2030.

‍

Is the use of AI for climate protection also worthwhile in SMEs?

Yes, provided that specific action is taken. Many SMEs can achieve significant savings with relatively simple AI applications — for example in energy monitoring, route planning or production optimization. It is important to start small, measure effects and embed AI into an overarching sustainability strategy.

‍

What can I do as a company if I have no influence on the data centers?

Even then, you have levers: choosing green providers, minimizing unnecessary inquiries, using more efficient models, using regional data centers with a high share of renewable energy and offsetting unavoidable emissions. You can also formulate transparency and sustainability requirements for AI providers in tenders.

‍

‍

‍

AI and climate: Conclusion and outlook

‍

AI and climate will move even closer together in the next few years. On the one hand, because AI can hardly be operated in a climate-friendly manner without sustainable infrastructure. On the other hand, because many climate goals are barely achievable without data-driven, intelligent systems — for example in the energy system, in mobility or in industry.

‍

It is crucial that companies see AI not only as an efficiency machine, but as an integral part of their sustainability strategy. Anyone who defines clear goals, prioritizes climate-positive use cases, integrates green compute criteria into tool selection and creates transparency about their own AI footprint can bring AI and climate into a constructive balance.

‍

The KI Company helps companies to implement exactly that: from identifying meaningful climate protection use cases to selecting suitable platforms to measuring and optimizing the ecological impact of AI projects. If you would like to strategically combine AI and climate in your own company, we would be happy to advise you without obligation.