Universities are fundamentally rethinking artificial intelligence's relationship with climate change, moving beyond the narrative that AI is simply an environmental burden. Instead of asking whether AI can be made sustainable, researchers at institutions like Duke University and Simon Fraser University are asking a deeper question: can AI become a coordination technology that helps humanity perceive and respond to ecological crises at the speed and scale of living systems themselves?

Why Universities Are Treating AI as a Climate Opportunity, Not Just a Problem

The contradiction seems obvious. The International Energy Agency (IEA) projects that global data center electricity consumption could more than double to 945 terawatt-hours by 2030, with artificial intelligence as the most important driver. Yet researchers argue that AI's true value lies not in whether the technology itself is "green," but in whether it can shorten the distance between an ecological signal and an institutional response.

At Duke University, leaders are pursuing this vision on two fronts simultaneously. The university is expanding its AI infrastructure with a small GPU (graphics processing unit) center expected to open in 2027, designed to minimize power and water consumption and carbon emissions through energy-efficient practices. Meanwhile, Duke researchers are using AI to improve climate modeling, strengthen weather forecasting, protect oceans and modernize energy systems.

"Climate change is one of the defining challenges of our time, and AI can be an important part of the solution, but only if we develop it responsibly. Duke's commitment is to lead in both directions: advancing AI that helps society understand and respond to a changing planet while pioneering the energy-efficient, ethical, and equitable technologies needed to make AI itself more sustainable," said Toddi Steelman, Duke's vice president and vice provost for climate and sustainability.

Toddi Steelman, Vice President and Vice Provost for Climate and Sustainability at Duke University

How Are Universities Making AI Hardware More Efficient?

The energy demands of AI data centers represent a genuine challenge. Large data center power demands are expected to rise about 130 percent by the end of this decade, with approximately one-half of that load growth driven by AI. But engineers are developing novel approaches to reduce consumption at the hardware level.

• Brain-Inspired Semiconductors: Electrical and Computer Engineering Professor Tania Roy at Duke is creating neuromorphic, or brain-inspired, semiconductor devices capable of performing AI tasks on their own, without the need of a data center. Her team is developing tiny experimental circuits that mimic the efficiency of the human brain, allowing future smart devices to process information locally rather than sending every request to energy-intensive server farms.
• Advanced Memory Technologies: Helen Li and Yiran Chen at Duke are working on magnetic random access memory approaches to reduce the water and energy consumption of AI. Their award-winning work is contributing to future and more efficient AI systems.
• AI-Enabled Battery Research: At Simon Fraser University, Tina Shoa is developing an AI-enabled battery research platform that can monitor battery health in real time, detect problems early, and improve the safety, performance and lifespan of next-generation energy storage technologies.

Chen and Roy are among researchers recently awarded a project funded by the Department of Energy to develop robotic AI processing hardware that is significantly faster and more efficient on local hardware, with the intent to reduce the amount of energy data sent to and from data centers.

What Can AI Actually Do to Help Predict and Prevent Climate Disasters?

Beyond hardware efficiency, universities are deploying AI to solve one of climate science's most pressing problems: the lag between observation and response. Governments often act after forests burn, fisheries collapse, or floods displace communities. AI introduces a different sequence: observation, prediction, prevention, adaptation and learning.

Weather forecasting offers a concrete example. GenCast, an AI system, produced 15-day probabilistic weather forecasts that outperformed the European Centre for Medium-Range Weather Forecasts' leading operational ensemble system across most variables and lead times tested. Such forecasts can support evacuation, agriculture, renewable-energy management and disaster preparedness.

Flood prediction carries even greater social consequence. Many vulnerable communities are located in river basins with few gauges and limited forecasting capacity. A global study found that an AI system could generate useful forecasts for ungauged watersheds and, in many cases, match or exceed the reliability of a major conventional global forecasting system. At Simon Fraser University, Mengxin Pan is establishing the new Climate and Weather Extremes Laboratory that uses advanced computing and AI to improve predictions of extreme weather and help communities prepare for the impacts of climate change.

Climate scientists at Duke are also using AI to recognize complex patterns hidden within enormous datasets. In the Nicholas School of the Environment, Professor Shineng Hu applies machine learning and deep learning to improve forecasts of climate extremes such as El Niño events and climate-related societal impacts, including malaria outbreaks in South America.

How Is AI Helping Scientists Monitor Marine Life and Coastal Ecosystems?

As climate change rapidly reshapes marine habitats, Duke's Marine Robotics and Remote Sensing Lab is combining drones, satellites and artificial intelligence to help scientists observe wildlife and coastal ecosystems at the speed and scale required for effective conservation. The lab uses AI to turn imagery collected by drones and satellites into practical information for marine conservation, addressing a fundamental bottleneck in climate and biodiversity science.

"In the past, scientists didn't have enough data. Now the script is flipping, and we are often data rich but without enough time to go through all the data. AI can help us analyze our data in 20 percent of the time, and we can rapidly see the results," said David Johnston, director of Duke's Marine Robotics and Remote Sensing Lab.

David Johnston, Director of Duke Marine Robotics and Remote Sensing Lab

The lab has developed systems that can detect animals, estimate population size, identify species, measure body dimensions, map habitat and document environmental change through studies of whales, harbor seals, sea turtles and seabirds, as well as oyster reefs, wetlands and coastal forests. The aim is to reduce analytical bottlenecks so that researchers and managers can monitor vulnerable species more frequently across larger areas and in places that are dangerous or difficult to reach.

What Does This Shift Mean for How We Understand Ecological Intelligence?

The deeper philosophical shift happening at universities goes beyond specific applications. Researchers are asking whether AI can help move society from "industrial intelligence" to "ecological intelligence". Industrial economies push consequences out of sight: atmospheric carbon, extraction at sea, remote biodiversity loss, and exported electronic waste. AI, combined with satellite imagery and sensors, can bring some of these hidden activities into public view.

A 2024 Nature study led by Global Fishing Watch analyzed two petabytes of satellite data to map industrial activity at sea. It found that 72 to 76 percent of industrial fishing vessels were absent from public tracking systems, with much of the untracked activity concentrated around South and Southeast Asia and Africa. This does not end illegal or unsustainable fishing, but it changes the politics of invisibility. What was once denied because it was hard to observe becomes available for investigation and public scrutiny.

However, researchers emphasize that data cannot simply "speak for nature." Sensors reflect what humans choose to measure, and models inherit the priorities and gaps embedded in their data. Ecological intelligence must integrate computational visibility with field science, community memory and Indigenous knowledge.

The investments being made across universities reflect a growing recognition that AI's environmental impact depends entirely on how it is developed and deployed. Duke's Nicholas Institute for Energy, Environment and Sustainability released a first-of-its-kind analysis finding that taking advantage of load flexibility could enable the U.S. power system to more quickly absorb new AI-driven demand while mitigating the immediate need for costly expansion of grid capacity. This suggests that the infrastructure challenge, while real, may be solvable through smarter grid management and hardware innovation working in tandem.