AI for a Greener Future: Its Power is in Our Hands

Can AI guide us toward a more sustainable future, or is it exacerbating global energy and climate challenges? 

This critical question was recently posed to a panel of sustainability and AI experts from Columbia University, Deloitte, and the Wilson Center at NVIDIA GTC 2025. 

In a packed room moderated by Josh Parker, senior director of Corporate Sustainability at NVIDIA, the discussion explored AI’s meteoric rise and its potential to help or hinder climate mitigation efforts. 

A well-echoed global concern is whether the explosive growth in AI technologies and the significant energy demands of data storage, computational power, and low-latency processing are fueling unsustainable power consumption, especially as these technologies continue to expand. 

While AI is already changing sectors like healthcare and finance, it’s also firmly embedded in people’s daily lives through tools like voice assistants (Siri or Alexa), recommender systems (Netflix or Spotify), and cloud-based services (Google Cloud and AWS). As it continues to evolve and integrate into modern life, the experts offered some surprising optimism, albeit with some significant caveats, that could guide how AI can help create a greener future.  

The AI energy footprint: manageable but growing

According to a newly released report from the ​International Energy Agency, data centers accounted for about 1.5% of global electricity use in 2024 and are projected to consume 6% by 2035. 

“There is a misconception that AI is a major contributor to global energy consumption, but the numbers tell a more nuanced story,” said Bernhard Lorentz, managing partner and global consulting sustainability and climate strategy leader at Deloitte.

Lorentz explained that while AI applications continue to expand and drive up power usage, their contribution remains relatively small compared to other sectors such as manufacturing and transportation. 

The core challenge, he stressed, lies in the unexpected rapid growth of AI in regions where the infrastructure isn’t equipped to handle it, causing spikes in energy demand. Strained local power grids are less efficient and less stable. They also rely more heavily on fossil fuel sources, which increases carbon emissions and air pollution. 

Despite this, the panelists agreed that the energy footprint is within manageable limits. Strategic decisions about where to build data centers and the energy sources to power them will be paramount to preventing strain on local energy grids.

David Sandalow, inaugural fellow at the Center on Global Energy Policy and co-director of the Energy and Environment Concentration at the School of International and Public Affairs, Columbia University, also pointed out that with each new generation of AI chips, there’s a notable increase in energy efficiency, where the technology achieves more with less energy.

“The energy efficiency for accelerated computing and AI for inference is 100,000 times better than it was, so something that used to take 100,000 terawatt hours now [would take] just one terawatt hour in terms of energy consumption for inference,” Parker added.

If AI can continue to guide this trend, its long-term value could outweigh potential downsides.  

Bridging the gap: AI and climate policy

One of the biggest challenges is the knowledge gap between AI experts and climate policy professionals, along with policies currently guiding AI’s growth.  

Lauren Risi, program director at the Wilson Center’s Environmental Change and Security Program, noted that policymakers often struggle to understand AI’s capabilities and limitations in the context of climate change. This disconnect makes creating effective regulations while fully leveraging AI for sustainability challenging. 

“These are two very complicated sectors, and you put them together and you have this really terrible game of whack-a-mole where you get certainty around one issue, but there are five other issues that pop up,” Risi said. 

Creating a shared understanding and language between AI and climate experts is essential for driving meaningful change, Risi emphasized. As AI continues to evolve, it’s crucial that governments, companies, and researchers collaborate closely to ensure that AI is being applied in ways that are both effective and sustainable. 

Optimizing energy and accelerating innovation

The conversation turned more positive as the panel discussed the capacity of AI to drive sustainable solutions. David Sandalow outlined how AI could improve energy systems and accelerate innovations in clean technologies. 

AI modeling tools are enhancing weather predictions and benefiting renewable energy sectors like solar and wind. More accurate forecasts can help farms optimize solar panel angles and orientation or adjust wind turbines for changing conditions, maximizing energy production. Smart grid management, powered by sensors and machine learning tools, can also help safely increase the amount of ‌renewable power put through a transmission line, facilitating the integration of renewable energy into the grid.

Turning to material science discovery, Sandalow emphasized how AI can help researchers discover and test new materials, including those for energy storage and carbon capture technologies. 

“I think what I’m most excited about is the transformational opportunity with AI, and for me, the most exciting areas might be materials innovation,” Sandalow said. 

Drawing a parallel with Thomas Edison, who took more than a year to develop the modern light bulb by testing dozens of different materials and running electricity through them to see how much light and heat would be produced, Sandalow explained that AI’s ability to simulate and analyze material properties could lead to breakthroughs. These advancements could revolutionize industries like batteries, lightweight materials, and other energy-efficient technologies.

“Today we can simulate a million of those interactions in a second,” Sandalow said. “We can even test materials that don’t yet exist,” he added.

AI’s role in sustainability

A key closing moment came from Sandalow’s final comments.

“I’ve been asked again and again… if AI can increase greenhouse gas emissions or decrease greenhouse gas emissions… I have a very confident answer I absolutely can tell you, which is nobody knows, and it depends on us. AI has the potential to decrease greenhouse gas emissions that contribute to climate change solutions, but it won’t necessarily do that unless we mobilize and make that happen.”

One of the biggest takeaways from the session was that it’s not enough for AI to have the potential to drive sustainability, and that there isn’t one approach for fixing everything. 

“AI is a potential enabler of some of these huge moon shots that we need to try to innovate our way out of challenges that we may not be able to solve with policy,” Parker said. 

The challenge is mobilizing AI’s capabilities in ways that align with climate goals, and this requires ‌concerted efforts from policymakers, businesses, and innovators to ensure AI is used effectively and responsibly. 

“We’re measuring deforestation rates, we’re quantifying carbon storage, AI is 10,000 times faster at analyzing changes in glaciers than humans are, but the measurements and the predictability provided through AI are only as good as our ability to act on it,” Risi said. 

As AI becomes increasingly recognized as a critical tool in addressing climate change, especially when combined with strong climate policies, it has the potential to optimize processes, reduce waste, and drive efficiency, making it an indispensable part of the solution.

Watch the AI, Energy, and Climate: Driving Sustainability and Energy Security session.