As artificial intelligence is changing the world, it’s also causing dramatic growth in demand for energy and other resources, with serious implications for humanity and the planet. But could AI also play an essential role in solving our energy challenges?

“The scale of the demand and the complexity of AI infrastructure is an order of magnitude greater than it was for previous generations of digital infrastructure,” says Mohammad Hajiesmaili (pictured above, left), associate professor in UMass Amherst’s Manning College of Information & Computer Sciences, who was named one of Popular Science’s “Brilliant 10” for his work to reduce computing’s carbon footprint. 

UMass faculty have played a leadership role in computing sustainability research for over two decades. Prashant Shenoy, distinguished professor of computer science, was the founding director of the Association for Computing Machinery’s Special Interest Group on Energy Systems and Informatics (ACM SIGEnergy), which today is led by David Irwin, professor in UMass’s Riccio College of Engineering. In June 2026, the group hosted its first annual ACM Sustainability Week in Banff, Canada, to convene researchers working to apply computational techniques to optimize energy and carbon emissions from different perspectives.

Today, as AI is supersizing demand for resources, UMass faculty are leading the charge in developing innovative solutions that capitalize on AI’s robust capabilities.

Image

In 2024, a group of UMass Amherst computer science and engineering faculty, led by Shenoy, received a $12 million, five-year National Science Foundation Expeditions grant to develop the field of computational decarbonization. Working with collaborators at satellite sites—including the University of Chicago, UCLA, MIT, Carnegie Mellon University, and the University of Wisconsin—they are applying computational approaches to automate decarbonization across the electrical grid, the built environment, transportation, and computing itself, which is responsible for a rapidly growing share of the world’s energy usage. 

Also, in spring 2026, UMass awarded a Strategic Partnerships to Advance Research and Creative activity (SPARC) grant to a group of researchers, led by Hajiesmaili, with a goal of establishing a large-scale research and workforce development initiative at UMass in sustainable computing and AI. 

“Both projects have a strong focus on AI and data centers, where the growth of computing is increasingly tied to questions of energy, infrastructure, and long-term societal impact,” says Shenoy. “These projects will develop the research foundations, testbeds, and workforce needed to help future AI systems scale in ways that are reliable, efficient, and sustainable.”
 

Above, in foreground, Prashant Shenoy (left) and David Irwin (right).

Reducing Computing’s Carbon Footprint: From Micro Edges to Outer Space

In his spring 2026 Distinguished Faculty Lecture on “Building a Faster, Smarter, and Greener Internet,” Ramesh Sitaraman (pictured at top of page, right), distinguished professor of computer science, issued a warning: “The future internet holds extraordinary promise—and extraordinary risk. As scientists, we bear a special responsibility to shape which of those futures prevails.”

Sitaraman has researched the internet for three decades, pioneering contributions to content delivery, edge computing, internet performance, and distributed systems. For the past 15 years, his research has focused on reducing the high energy usage required by large computing systems. Though he initially pursued overall energy efficiency, he later shifted his attention to reducing carbon emissions.

“Rather than reducing overall energy use, we can make progress on sustainability by shifting to less carbon-intensive energy sources,” he explains. This led him to develop a tool called CarbonCast, which—similar to a weather forecast—can predict the carbon intensity of a given power source over the next 96 hours. With this information, he aims to develop algorithms to move computational workload from locations with fewer renewable energy sources to those with more in order to reduce overall carbon emissions.

Sitaraman is also studying ways to redesign the infrastructure behind the internet to be more sustainable. Currently, the internet relies on massive data centers that each host hundreds of servers. Powering such large data centers with renewable energy alone is unrealistic, which is why Sitaraman is developing strategies to build small collections of servers, known as “micro edges,” that can be easily powered by renewable sources such as solar and wind energy. While such small data centers may be less reliable overall than the current large, heavily fortified data centers, Sitaraman says, “We’re thinking about how we can create clever algorithms and systems so that even if, say, 10 or 20 percent of micro edges are not functional at a given time, the overall system would still work reliably.”

Looking to the future, Sitaraman is also exploring more far-flung possibilities to improve sustainability, including deploying data centers on Low Earth Orbit (LEO) satellites. This idea is attractive for a few reasons—including plenty of access to solar power and no water consumed for cooling the servers—but also poses special challenges. Specifically, LEO satellites are not stationary but move quickly, orbiting the Earth about once an hour, requiring smart algorithms to move data efficiently between servers that are in constant motion. Furthermore, the satellite’s launch, repair, and replacement lifecycle causes carbon emissions that need to be accounted for and reduced for this approach to be sustainable. 
 

AI and Clean Energy: A Match Made in Heaven

Image

AI-related energy use has grown by staggering amounts, with no signs of slowing. Meeting this demand while continuing to provide electricity to the rest of society will require reimagining the design of computing and energy infrastructure to be scalable, reliable, and sustainable.

This is a daunting challenge, but there are factors working in our favor. “AI infrastructure has a flexibility that, if utilized, can be perfectly paired with the recent expansion of clean energy,” Hajiesmaili explains. “One of the biggest issues with clean energy is that it’s intermittent; it depends on the availability of sun and wind, for example. But AI infrastructure can be designed to adapt to signals from the grid.”

Through experimental research on time-shifting energy usage, for which Shenoy won an ACM e-Energy 2024 Test of Time Award, he is developing algorithms that account for grid conditions—such as cost, supply–demand balance, and availability of green energy—to optimize the timing of AI workloads. He has worked with the Massachusetts Green High Performance Computing Center (MGHPCC) since its beginnings in the 2010s to power computing with green energy. To facilitate experimental research, Shenoy and Irwin built a micro data center powered by solar with battery support at the MGHPCC site. “Our solar and battery-powered micro data center has enabled us to experimentally test a range of carbon-aware algorithms under real-world conditions,” says Irwin.

Hajiesmaili adds that ensuring enough electricity for all will depend not only on adding new power capacity, but also on using the grid more intelligently. Many AI workloads can tolerate carefully managed pauses, delays, or small reductions in service quality, allowing data centers to reduce demand during periods of grid stress and shift work to times or places where power is cleaner, cheaper, or more available. Managed properly, this flexibility could make data centers part of the energy affordability solution rather than simply another source of pressure on ratepayers. It will also require public policy and planning processes that support this kind of demand response while addressing the siting, permitting, and community impacts associated with building large-scale data centers.

Despite the challenges ahead, Sitaraman is sanguine about the future. He cautions against going to extremes—either putting the brakes entirely on AI or rejecting any sort of regulation. 

“I think there must be a conversation so that we can shape this technology in ways that benefit society. We must consider the social costs and ensure we are taking care of the people who are impacted the most,” he says.

This story was originally published in June 2026.