The global economy faces unprecedented environmental crises, including climate change, biodiversity loss and widespread pollution. Achieving a transition to net-zero emissions requires fundamental systemic changes and massive investments in low-carbon infrastructure and technologies, particularly in emerging markets and developing economies (EMDEs). It is estimated that global climate investments will amount to at least $4 trillion by 2030. This “green investment push” represents not only an opportunity for incremental improvements, but a chance for systemic transformation, which can launch a new growth story, supporting sustainable, inclusive and resilient economic growth.

Artificial intelligence (AI) is ideally positioned to accelerate this transition. As a general-purpose technology, AI can increase the speed, efficiency, and effectiveness of innovation processes and the deployment of capital. However, despite its potential, there is limited research on its specific impact on the low-carbon transition and its macro-level effects. A new study identifies five key areas where AI can significantly support the climate transformation:

• Transformation of complex systems: Decarbonizing the global economy requires radical changes in key systems such as cities, land, transport and energy. AI can optimize grid management, improve the integration of renewable energy sources (RES) and predict investment risks in low-carbon projects, thereby streamlining financing in EMDEs.
• Technology innovation and resource efficiency: Nearly half of the emissions reductions needed to reach net-zero by 2050 will come from technologies currently in the prototype or demonstration phase. AI can accelerate the discovery of new materials (e.g. Google DeepMind GNoME has identified millions of theoretical crystal structures) and improve resource efficiency in industry, logistics and recycling (e.g. Amazon in packaging optimization, GreyParrot in waste sorting).
• Influencing behavior and habits: Changes in lifestyle and consumer behavior can reduce greenhouse gas emissions by 40-70 % by 2050. AI can provide personalized recommendations for low-carbon options, optimize energy use in homes (e.g. Google Nest, Oracle Opower), and help reduce food waste (e.g. Winnow Vision). In transportation, Google Maps already offers fuel-efficient routes.
• Modeling climate systems and policy interventions: AI can process large data sets and simulate complex scenarios in real time, improving the accuracy of climate models and weather forecasts (e.g. IceNet for sea ice). It can also help design and monitor the effectiveness of climate policies (e.g. Climate Policy Radar) and develop economic models that include metrics “beyond GDP.”
• Adaptation and resilience management: With the increasing frequency of climate disasters, the ability to predict risks and adapt is key. AI is already improving early warning systems for extreme events (e.g. Google FloodHub for floods), helping to prevent damage and save lives. It is also supporting long-term resilience through large-scale simulations, for example, in tracking biodiversity loss after wildfires.

Quantifying the potential for AI emissions reductions

Study, which focused on three key sectors – energy, meat and dairy, and light road vehicles – which together contribute to almost half of global emissions, estimates the significant potential of AI to reduce greenhouse gas emissions. AI could reduce global emissions by 3.2-5.4 GtCO2e per year by 2035 from these three sectors only.

• IN energy AI can optimize grid management and increase the efficiency of solar and wind power, which would lead to a reduction in emissions of 1.8 GtCO2e per year.
• In the sector meat and dairy products AI is expected to increase the adoption of alternative proteins (APs) by improving their taste and texture and reducing production costs. This could lead to emission savings of up to 1.7–3.0 GtCO2e per year in an ambitious scenario.
• In the sector light road vehicles AI-enhanced shared mobility and improvements in the availability of electric vehicles (EVs) (e.g. cheaper batteries, optimal placement of charging stations) could save 0.5–0.6 GtCO2e annually.

These estimated emission reductions are important because they more than exceed the estimated increase in emissions from global energy consumption of data centers and AI activities (0.4-1.6 GtCO2e). This strongly supports the argument that using AI to address the climate threat is not only important, but downright necessary.

Limitations and the role of the state

Despite its promising results, the study has limitations. It focuses on just three sectors and does not quantify the dynamic impact that could arise from mutually reinforcing changes in different economic systems. It also does not include the potential of AI to accelerate capital deployment or support better policymaking. It also does not take into account so-called “rebound effects,” where increased efficiency can lead to higher consumption.

It is crucial to realize that Leaving markets to determine AI applications and governance can be risky. The role of the “active state” is critical to ensuring that AI is deployed to accelerate the low-carbon transition in a fair and sustainable manner. Governments must create enabling conditions for AI deployment, provide financial incentives for research and development, and ensure that AI applications are directed towards public good and high-impact areas. In addition, it is important to regulate AI in a way that minimizes its environmental footprint (e.g., supporting renewable energy for data centers) and avoid widening inequalities between the global North and South.

AI represents an unprecedented opportunity for a net-zero transition, with its potential emissions savings far exceeding its own energy intensity. However, strategic management and active intervention by governments are essential to fully realize this potential. Spring

Glossary of key terms

• Artificial Intelligence (AI): A generally applicable technology that can accelerate the process of systemic transformation by increasing the speed, efficiency, and effectiveness of innovation processes and capital deployment.
• Transition to net zero emissions: A global effort to reduce greenhouse gas emissions to a minimum, with any remaining emissions removed from the atmosphere to achieve an overall balance.
• Greenhouse gas emissions (GHG): Gases in the atmosphere that absorb and radiate heat, causing the greenhouse effect and contributing to global warming.
• Sigmoid curve (S-curve) patterns: A non-linear trajectory used to model technology adoption in five stages: concept, solution development, niche market, mass market, and late market.
• Affordability: The cost of the solution compared to alternatives, which is key to user or investor adoption by ensuring financial competitiveness.
• Attractiveness: The appeal of the solution to users, including performance, convenience, and other benefits that increase desirability.
• Accessibility: The ease with which users can access the solution, including distribution, supply chain stability, and system integration.
• Computational General Equilibrium (CGE) Modeling: An economic model used to analyze interactions between different sectors of the economy and to assess large-scale societal impacts.
• Bottom-up methodology: An approach that analyzes the specific impact potential of AI within individual sectors and their applications.
• Reflection effects: Unintended or additional changes in behavior or consumption that result from efficiency gains that can potentially reduce or increase net emission reductions.
• Digital twins: Virtual representations of physical objects, systems, or processes that are used to simulate and predict real-world behavior and interactions, such as NVIDIA Earth-2 for weather forecasting.
• Alternative proteins (APs): Food products that replace traditional meat and dairy products and typically have a lower carbon footprint.
• Distributed Energy Resources (DERs): Small, modular energy sources that are decentralized and located close to the point of consumption, such as electric vehicles and energy storage systems.
• Active state: A concept that highlights the key role of government in ensuring that AI is deployed fairly and sustainably to accelerate the low-carbon transition, overcoming the limitations of market forces.