Transistors, the building blocks of modern electronics, are typically made of silicon. Because it’s a semiconductor, this material can control the flow of electricity in a circuit. But silicon has fundamental physical limits that restrict how compact and energy-efficient a transistor can be.

MIT researchers have now replaced silicon with a magnetic semiconductor, creating a magnetic transistor that could enable smaller, faster, and more energy-efficient circuits. The material’s magnetism strongly influences its electronic behavior, leading to more efficient control of the flow of electricity. 

The team used a novel magnetic material and an optimization process that reduces the material’s defects, which boosts the transistor’s performance.

The material’s unique magnetic properties also allow for transistors with built-in memory, which would simplify circuit design and unlock new applications for high-performance electronics.

“People have known about magnets for thousands of years, but there are very limited ways to incorporate magnetism into electronics. We have shown a new way to efficiently utilize magnetism that opens up a lot of possibilities for future applications and research,” says Chung-Tao Chou, an MIT graduate student in the departments of Electrical Engineering and Computer Science (EECS) and Physics, and co-lead author of a paper on this advance.

Chou is joined on the paper by co-lead author Eugene Park, a graduate student in the Department of Materials Science and Engineering (DMSE); Julian Klein, a DMSE research scientist; Josep Ingla-Aynes, a postdoc in the MIT Plasma Science and Fusion Center; Jagadeesh S. Moodera, a senior research scientist in the Department of Physics; and senior authors Frances Ross, TDK Professor in DMSE; and Luqiao Liu, an associate professor in EECS, and a member of the Research Laboratory of Electronics; as well as others at the University of Chemistry and Technology in Prague. The paper appears today in Physical Review Letters.

Overcoming the limits

In an electronic device, silicon semiconductor transistors act like tiny light switches that turn a circuit on and off, or amplify weak signals in a communication system. They do this using a small input voltage.

But a fundamental physical limit of silicon semiconductors prevents a transistor from operating below a certain voltage, which hinders its energy efficiency.

To make more efficient electronics, researchers have spent decades working toward magnetic transistors that utilize electron spin to control the flow of electricity. Electron spin is a fundamental property that enables electrons to behave like tiny magnets.

So far, scientists have mostly been limited to using certain magnetic materials. These lack the favorable electronic properties of semiconductors, constraining device performance.

“In this work, we combine magnetism and semiconductor physics to realize useful spintronic devices,” Liu says.

The researchers replace the silicon in the surface layer of a transistor with chromium sulfur bromide, a two-dimensional material that acts as a magnetic semiconductor.

Due to the material’s structure, researchers can switch between two magnetic states very cleanly. This makes it ideal for use in a transistor that smoothly switches between “on” and “off.”

“One of the biggest challenges we faced was finding the right material. We tried many other materials that didn’t work,” Chou says.

They discovered that changing these magnetic states modifies the material’s electronic properties, enabling low-energy operation. And unlike many other 2D materials, chromium sulfur bromide remains stable in air.

To make a transistor, the researchers pattern electrodes onto a silicon substrate, then carefully align and transfer the 2D material on top. They use tape to pick up a tiny piece of material, only a few tens of nanometers thick, and place it onto the substrate.

“A lot of researchers will use solvents or glue to do the transfer, but transistors require a very clean surface. We eliminate all those risks by simplifying this step,” Chou says.

Leveraging magnetism

This lack of contamination enables their device to outperform existing magnetic transistors. Most others can only create a weak magnetic effect, changing the flow of current by a few percent or less. Their new transistor can switch or amplify the electric current by a factor of 10.

They use an external magnetic field to change the magnetic state of the material, switching the transistor using significantly less energy than would usually be required.

The material also allows them to control the magnetic states with electric current. This is important because engineers cannot apply magnetic fields to individual transistors in an electronic device. They need to control each one electrically.

The material’s magnetic properties could also enable transistors with built-in memory, simplifying the design of logic or memory circuits.

A typical memory device has a magnetic cell to store information and a transistor to read it out. Their method can combine both into one magnetic transistor.

“Now, not only are transistors turning on and off, they are also remembering information. And because we can switch the transistor with greater magnitude, the signal is much stronger so we can read out the information faster, and in a much more reliable way,” Liu says.

Building on this demonstration, the researchers plan to further study the use of electrical current to control the device. They are also working to make their method scalable so they can fabricate arrays of transistors.

This research was supported, in part, by the Semiconductor Research Corporation, the U.S. Defense Advanced Research Projects Agency (DARPA), the U.S. National Science Foundation (NSF), the U.S. Department of Energy, the U.S. Army Research Office, and the Czech Ministry of Education, Youth, and Sports. The work was partially carried out at the MIT.nano facilities.

Large language models (LLMs) sometimes learn the wrong lessons, according to an MIT study.

Rather than answering a query based on domain knowledge, an LLM could respond by leveraging grammatical patterns it learned during training. This can cause a model to fail unexpectedly when deployed on new tasks.

The researchers found that models can mistakenly link certain sentence patterns to specific topics, so an LLM might give a convincing answer by recognizing familiar phrasing instead of understanding the question.

Their experiments showed that even the most powerful LLMs can make this mistake.

This shortcoming could reduce the reliability of LLMs that perform tasks like handling customer inquiries, summarizing clinical notes, and generating financial reports.

It could also have safety risks. A nefarious actor could exploit this to trick LLMs into producing harmful content, even when the models have safeguards to prevent such responses.

After identifying this phenomenon and exploring its implications, the researchers developed a benchmarking procedure to evaluate a model’s reliance on these incorrect correlations. The procedure could help developers mitigate the problem before deploying LLMs.

“This is a byproduct of how we train models, but models are now used in practice in safety-critical domains far beyond the tasks that created these syntactic failure modes. If you’re not familiar with model training as an end-user, this is likely to be unexpected,” says Marzyeh Ghassemi, an associate professor in the MIT Department of Electrical Engineering and Computer Science (EECS), a member of the MIT Institute of Medical Engineering Sciences and the Laboratory for Information and Decision Systems, and the senior author of the study.

Ghassemi is joined by co-lead authors Chantal Shaib, a graduate student at Northeastern University and visiting student at MIT; and Vinith Suriyakumar, an MIT graduate student; as well as Levent Sagun, a research scientist at Meta; and Byron Wallace, the Sy and Laurie Sternberg Interdisciplinary Associate Professor and associate dean of research at Northeastern University’s Khoury College of Computer Sciences. A paper describing the work will be presented at the Conference on Neural Information Processing Systems.

Stuck on syntax

LLMs are trained on a massive amount of text from the internet. During this training process, the model learns to understand the relationships between words and phrases — knowledge it uses later when responding to queries.

In prior work, the researchers found that LLMs pick up patterns in the parts of speech that frequently appear together in training data. They call these part-of-speech patterns “syntactic templates.”

LLMs need this understanding of syntax, along with semantic knowledge, to answer questions in a particular domain.

“In the news domain, for instance, there is a particular style of writing. So, not only is the model learning the semantics, it is also learning the underlying structure of how sentences should be put together to follow a specific style for that domain,” Shaib explains.   

But in this research, they determined that LLMs learn to associate these syntactic templates with specific domains. The model may incorrectly rely solely on this learned association when answering questions, rather than on an understanding of the query and subject matter.

For instance, an LLM might learn that a question like “Where is Paris located?” is structured as adverb/verb/proper noun/verb. If there are many examples of sentence construction in the model’s training data, the LLM may associate that syntactic template with questions about countries.

So, if the model is given a new question with the same grammatical structure but nonsense words, like “Quickly sit Paris clouded?” it might answer “France” even though that answer makes no sense.

“This is an overlooked type of association that the model learns in order to answer questions correctly. We should be paying closer attention to not only the semantics but the syntax of the data we use to train our models,” Shaib says.

Missing the meaning

The researchers tested this phenomenon by designing synthetic experiments in which only one syntactic template appeared in the model’s training data for each domain. They tested the models by substituting words with synonyms, antonyms, or random words, but kept the underlying syntax the same.

In each instance, they found that LLMs often still responded with the correct answer, even when the question was complete nonsense.

When they restructured the same question using a new part-of-speech pattern, the LLMs often failed to give the correct response, even though the underlying meaning of the question remained the same.

They used this approach to test pre-trained LLMs like GPT-4 and Llama, and found that this same learned behavior significantly lowered their performance.

Curious about the broader implications of these findings, the researchers studied whether someone could exploit this phenomenon to elicit harmful responses from an LLM that has been deliberately trained to refuse such requests.

They found that, by phrasing the question using a syntactic template the model associates with a “safe” dataset (one that doesn’t contain harmful information), they could trick the model into overriding its refusal policy and generating harmful content.

“From this work, it is clear to me that we need more robust defenses to address security vulnerabilities in LLMs. In this paper, we identified a new vulnerability that arises due to the way LLMs learn. So, we need to figure out new defenses based on how LLMs learn language, rather than just ad hoc solutions to different vulnerabilities,” Suriyakumar says.

While the researchers didn’t explore mitigation strategies in this work, they developed an automatic benchmarking technique one could use to evaluate an LLM’s reliance on this incorrect syntax-domain correlation. This new test could help developers proactively address this shortcoming in their models, reducing safety risks and improving performance.

In the future, the researchers want to study potential mitigation strategies, which could involve augmenting training data to provide a wider variety of syntactic templates. They are also interested in exploring this phenomenon in reasoning models, special types of LLMs designed to tackle multi-step tasks.

“I think this is a really creative angle to study failure modes of LLMs. This work highlights the importance of linguistic knowledge and analysis in LLM safety research, an aspect that hasn’t been at the center stage but clearly should be,” says Jessy Li, an associate professor at the University of Texas at Austin, who was not involved with this work.

This work is funded, in part, by a Bridgewater AIA Labs Fellowship, the National Science Foundation, the Gordon and Betty Moore Foundation, a Google Research Award, and Schmidt Sciences.

More than 300 people across academia and industry spilled into an auditorium to attend a BoltzGen seminar on Thursday, Oct. 30, hosted by the Abdul Latif Jameel Clinic for Machine Learning in Health (MIT Jameel Clinic). Headlining the event was MIT PhD student and BoltzGen’s first author Hannes Stärk, who had announced BoltzGen just a few days prior.

Building upon Boltz-2, an open-source biomolecular structure prediction model predicting protein binding affinity that made waves over the summer, BoltzGen (officially released on Sunday, Oct. 26.) is the first model of its kind to go a step further by generating novel protein binders that are ready to enter the drug discovery pipeline.

Three key innovations make this possible: first, BoltzGen’s ability to carry out a variety of tasks, unifying protein design and structure prediction while maintaining state-of-the-art performance. Next, BoltzGen’s built-in constraints are designed with feedback from wetlab collaborators to ensure the model creates functional proteins that don’t defy the laws of physics or chemistry. Lastly, a rigorous evaluation process tests the model on “undruggable” disease targets, pushing the limits of BoltzGen’s binder generation capabilities.

Most models used in industry or academia are capable of either structure prediction or protein design. Moreover, they’re limited to generating certain types of proteins that bind successfully to easy “targets.” Much like students responding to a test question that looks like their homework, as long as the training data looks similar to the target during binder design, the models often work. But existing methods are nearly always evaluated on targets for which structures with binders already exist, and end up faltering in performance when used on more challenging targets.

“There have been models trying to tackle binder design, but the problem is that these models are modality-specific,” Stärk points out. “A general model does not only mean that we can address more tasks. Additionally, we obtain a better model for the individual task since emulating physics is learned by example, and with a more general training scheme, we provide more such examples containing generalizable physical patterns.”

The BoltzGen researchers went out of their way to test BoltzGen on 26 targets, ranging from therapeutically relevant cases to ones explicitly chosen for their dissimilarity to the training data. 

This comprehensive validation process, which took place in eight wetlabs across academia and industry, demonstrates the model’s breadth and potential for breakthrough drug development.

Parabilis Medicines, one of the industry collaborators that tested BoltzGen in a wetlab setting, praised BoltzGen’s potential: “we feel that adopting BoltzGen into our existing Helicon peptide computational platform capabilities promises to accelerate our progress to deliver transformational drugs against major human diseases.”

While the open-source releases of Boltz-1, Boltz-2, and now BoltzGen (which was previewed at the 7th Molecular Machine Learning Conference on Oct. 22) bring new opportunities and transparency in drug development, they also signal that biotech and pharmaceutical industries may need to reevaluate their offerings. 

Amid the buzz for BoltzGen on the social media platform X, Justin Grace, a principal machine learning scientist at LabGenius, raised a question. “The private-to-open performance time lag for chat AI systems is [seven] months and falling,” Grace wrote in a post. “It looks to be even shorter in the protein space. How will binder-as-a-service co’s be able to [recoup] investment when we can just wait a few months for the free version?” 

For those in academia, BoltzGen represents an expansion and acceleration of scientific possibility. “A question that my students often ask me is, ‘where can AI change the therapeutics game?’” says senior co-author and MIT Professor Regina Barzilay, AI faculty lead for the Jameel Clinic and an affiliate of the Computer Science and Artificial Intelligence Laboratory (CSAIL). “Unless we identify undruggable targets and propose a solution, we won’t be changing the game,” she adds. “The emphasis here is on unsolved problems, which distinguishes Hannes’ work from others in the field.” 

Senior co-author Tommi Jaakkola, the Thomas Siebel Professor of Electrical Engineering and Computer Science who is affiliated with the Jameel Clinic and CSAIL, notes that "models such as BoltzGen that are released fully open-source enable broader community-wide efforts to accelerate drug design capabilities.”

Looking ahead, Stärk believes that the future of biomolecular design will be upended by AI models. “I want to build tools that help us manipulate biology to solve disease, or perform tasks with molecular machines that we have not even imagined yet,” he says. “I want to provide these tools and enable biologists to imagine things that they have not even thought of before.”

The U.S. military is mighty, formidable, and singular in influence, stationed in at least 128 overseas bases across 51 countries. Concealed beneath the United States’ biggest investment is a surprise: The military was responsible for the birth of an industry. Today, that industry is essential for its operations.

“If you think about it, logistics started as a military function,” says Chris Caplice, executive director of the MIT Center for Transportation and Logistics (CTL). “The idea of getting supplies, ammunition, food, all the material you need to the front line was the core of logistics, and really supply chain came out of that over the last decades or centuries.”

For Caplice and the leadership at MIT CTL, a collaboration with the U.S. military seemed inevitable. In 2006, MIT CTL launched the Military Fellows program, wherein three military logistics officers participate in the MIT Supply Chain Management master’s program. “The education goes two ways: One is that these people who have been in the service for more than 20 years step out of their silo and see all the research we’re doing that’s more focused on the private sector, and is cutting-edge. On the other side, you have students who may have never interacted with the military are able to learn from them,” reflects Caplice.

This year’s cohort holds 80 years of combined military service. It comprises Lukas Toth from the Army Reserve, Duston Mullen from the South Dakota Army National Guard, and Charles Greene from the Active Army. Though they work in different components of the U.S. Army, they all agree that their experience in the program so far has been humbling. 

“All of the MIT SCM students have strong academic backgrounds and are exceptionally better at math than us,” Toth laughs. “If you’re coming to this program, you’re sharp and you want to make a difference, not just in your life, but you want to make a difference in the world. Getting to sit in a room with 40 young people who want to make change happen and want to solve complex problems has been super rewarding.”

No strangers to being challenged, adversity is what called the fellows to become logistics officers in the first place. “It comes down to a quote I heard: Operations is easy, fighting the war is easy, but logistics is hard,” says Mullen. 

As logistics officers are responsible for everything from feeding soldiers to fixing trucks to warehousing and distribution, they must perform these functions at varying scales, and with varying threats to their operations. “Our work is: How do we enable the war fighter to be able to deliver when the nation requires? We’re looking at the supply chain and ensuring that we can deliver at the right time, right place, and in the right quantity with precision and accuracy,” says Greene. 

Although companies focus extensively on optimizing their supply chains for cost and efficiency, logistics officers in the military have an additional obstacle. “That last mile could be a contested mile, and the enemy gets a vote,” adds Toth. “At the end of the day, the civilian industry’s consumer has a product they want, and at the end of the day, our war fighters have a product that they want, but we have the added challenge of having to overcome a competitor who may go so far as to destroy us.”

Despite the fellows’ rich practical experience in the military, their academic experience still brings applicable use in terms of introduction to new technologies with which they hope to engage senior military leaders, insight into industry problem-solving to reduce overall military spending and influence decision-making, and, above all, communication. In the military, the stakes are higher than in the private sector, making communication rife with consequence. 

“This experience is helping us better communicate with industry and build an industry and logistics network so that if a challenge does come our country's way, we can better communicate with everybody to solve those challenges,” reflects Toth. 

In 1965, after completing his PhD in civil engineering at MIT, Professor Richard de Neufville joined the first class of White House Fellows, one of the nation’s most prestigious programs for leadership and public service, through which he spent an intensive year working full-time at the highest levels of government. Soon after, de Neufville joined the MIT faculty and led a steering committee that developed what would become the MIT Technology and Policy Program (TPP). TPP was approved in 1975 and launched in 1976 as an Institute-wide hub of education and research, and included a two-year, research-based master’s degree, with de Neufville serving as its founding chair.

This October, TPP held a symposium and celebration at MIT, marking TPP’s 50th year as an interdisciplinary effort focused on advancing the responsible leadership of technology through the integration of technical expertise and rigorous policy analysis in critical areas such as energy, the environment, security, innovation, and beyond.

As the 1988 “TPP Fact Book” stated: “The Technology and Policy Program educates men and women for leadership on the important technological issues confronting society. We prepare our graduates to excel in their technical fields, and to develop and implement effective strategies for dealing with the risks and opportunities associated with those technologies. This kind of education is vital to the future of our society.”

Now in its 50th year, TPP’s legacy of education, research, and impact has shaped more than 1,500 alumni who are among the most distinguished technology policy leaders across the world. TPP alumni often describe the program as life-changing and transformative — an educational experience that shaped their understanding of purpose, systems, and leadership in ways that continue to guide their careers throughout their lives. Today, over 50 TPP graduate students conduct research across the Institute on topics such as energy grid modeling, environmental protection, nuclear safety, industrial decarbonization, space system engineering and public policy, technoeconomic modeling of materials value chains, and governance of global digital systems and artificial intelligence.

Working to bring technically-informed and scientifically robust insights to technology policy is as urgent today as it was 50 years ago, says Christine Ortiz, Morris Cohen Professor of Materials Science and Engineering and the current director of TPP. “The role of technology policy is more essential than ever, helping to shape national and international priorities and underpinning societal and planetary well-being,” said Ortiz in her opening remarks. “Today’s symposium is convened with urgency amid a rapidly shifting landscape. We are situated here today at the epicenter of profound technological advancement, reaffirming our collective responsibility to ensure that innovation advances the well-being of humanity and the health of our planet.”

North stars and new routes

The TPP 50th Anniversary Symposium — North Stars and New Routes — held on Oct. 11, convened more than 630 participants from 30 countries, both in-person and virtually. The gathering brought together alumni, faculty, students, and global leaders to celebrate five decades of impact while exploring bold new directions for the future of technology and policy.

Over the course of seven thematic sessions and 45 speakers, the symposium offered a sweeping view of the current issues shaping the next era of technology policy. Discussions spanned a wide range of topics, including energy systems modeling, global environmental governance, ecologically neutral manufacturing, design of global digital systems, trust as national security infrastructure, the future of technology policy as a domain of scholarship, and the role of technology policy in the future of the research university.

The day opened with a dynamic panel examining the technical frontiers and possibilities of interactive energy systems modeling. Speakers highlighted the dual role of simulation tools as both advanced instruments for understanding decision outcomes and uncertainties, as well participatory platforms for engaging policymakers and stakeholders.

The next session, focused on global environmental governance, explored new approaches to planetary cooperation and emphasized how data-driven policy, equitable technology transfer, and accountability mechanisms can strengthen international climate action. Panelists called for adaptive and integrated governance frameworks that mirror the interconnectedness and complexity of the environmental systems they aim to protect.

In a session on ecologically neutral manufacturing, participants discussed advances in circular materials design and life-cycle modeling that reduce industrial emissions and resource intensity. Speakers underscored the importance of policies promoting reuse, recycling, and cleaner production — linking manufacturing innovation with both economic competitiveness and ecological resilience.

Turning to the design and governance of global digital systems, keynote speaker David Clark, senior research scientist at MIT’s Computer Science and Artificial Intelligence Laboratory and a pioneering architect of the internet, examined how the architecture of digital networks both reflects and shapes societal values, power, and accountability. He noted that the internet’s original open design — built for innovation and resilience — now faces pressing challenges of trust, privacy, and control. The next generation of digital infrastructure, he argued, must embed trust and accountability into its very foundations. The subsequent panel expanded on these themes, exploring how global digital ecosystems are influenced by the competing incentives of governments, corporations, and users. Speakers called for governance models that integrate technical, economic, and ethical considerations — emphasizing that true accountability depends not only on external regulation, but on embedding human values directly into the design of technology.

The theme of trust carried into the next discussion, with the focus on trust as infrastructure for security policy, where experts emphasized that national and global security must evolve to encompass cyber-trust, space governance, and technological resilience as essential infrastructures for stability in an era defined by AI and geopolitical complexity and uncertainty.

In the final session, which explored the role of technology policy in the future of the research university, panelists discussed how research institutions can strengthen their societal role by embedding technology policy and interdisciplinary scholarship into the institutional structure. Speakers emphasized the need for universities to evolve into more cohesive, outward-looking engines of policy innovation — coordinating existing centers of excellence, improving communication between research and government, and expanding educational pathways that integrate engineering, social science, and civic engagement.

Technology, policy, and power

In a keynote address, Senator Edward J. Markey, U.S. senator for Massachusetts, delivered a compelling call for moral and democratic leadership in governing the technologies shaping modern life. He warned that the rapid expansion of artificial intelligence and digital systems has outpaced the ethical and policy frameworks needed to protect society, declaring that “the privacy protections of all preceding generations have broken down.” Markey called for a renewed commitment to AI civil rights and accountability in the digital age, urging that technology must be harnessed as “a tool for connection, not addiction,” and developed to advance human dignity, fairness, and shared prosperity.

Framing technology as both a source of immense potential and a concentration of power, Markey argued that the defining question of our era is who controls that power, and to what end. He urged policymakers, researchers, and citizens alike to ensure that innovation strengthens democracy rather than undermines it. Closing on a note of determination and hope, Markey reminded the audience that technology policy is inseparable from human and planetary well-being: “Technology is power … the question is, who wields it and for what purpose. We must ensure it serves democracy, equality, and the future of our planet.”

New Institute-wide policy initiative announced

The symposium concluded with the announcement of an exciting new Institute-wide initiative, Policy@MIT, introduced by Maria Zuber, E.A. Griswold Professor of Geophysics and Presidential Advisor for Science and Technology Policy. Zuber described the effort as a bold and unifying step to synergize and amplify policy initiatives across MIT, strengthening the Institute’s capacity to inform evidence-based policymaking. 

Building upon the foundational work of TPP — within which the program will serve as a core pillar — Policy@MIT aims to connect MIT’s deep technical expertise with real-world policy challenges, foster collaboration across schools and disciplines, and train the next generation of leaders to ensure that science and technology continue to serve humanity and the planet.

Extending MIT TPP’s legacy of technology and policy leadership 

As MIT charts the next half-century of leadership at the intersection of technology, policy, and society, TPP continues to serve as a cornerstone of this mission. Operating within the MIT Institute for Data, Systems, and Society (IDSS), the MIT School of Engineering, and the MIT Schwarzman College of Computing, TPP distinctively engages and integrates state-of-the-art modeling, simulation, and analytical methods in information and decision systems, statistics and data science, and the computational social sciences, with a diverse range of foundational, emerging, and cross-disciplinary policy analysis methods. Sitting at the confluence of engineering, computer science, and the social sciences, TPP equips students and researchers to study some of the most important and complex emerging issues related to technology through systems thinking, technical rigor, and policy analysis.

Founding IDSS director Munther Dahleh, the William A. Coolidge Professor in Electrical Engineering and Computer Science, described this integration as cultivating the “trilingual student” — someone fluent in data and information, social reasoning, and a technical domain. “What we’re trying to produce in the TPP program,” he explained, “is the person who can navigate all three dimensions of a problem.”

Reflecting on TPP’s enduring mission, Ortiz concluded the symposium, “As we look ahead to the next 50 years, this is a pivotal moment for the Technology and Policy Program — both at MIT and globally. TPP holds tremendous potential for growth, translation, and impact as a leader in technology policy for the nation and the world.”

On a sunny, warm Sunday MIT students, staff, and faculty spread out across the fields of Hannan Healthy Foods in Lincoln, Massachusetts. Some of these volunteers pluck tomatoes from their vines in a patch a few hundred feet from the cars whizzing by on Route 117. Others squat in the shade cast by the greenhouse to snip chives. Still others slice heads of Napa cabbage from their roots in a bed nearer the woods. Everything being harvested today will wind up in Harvest Boxes, which will be sold at a pop-up farm stand the next day in the lobby of the Stata Center back on the MIT campus.

This initiative — a pilot collaboration between MIT’s Office of Sustainability (MITOS), MIT Anthropology, Hannan Healthy Foods, and the nascent MIT Farm student organization — sold six-pound boxes of fresh, organic produce to the MIT community for $10 per box — half off the typical wholesale price. The weekly farm stands ran from Sept. 15 through Oct. 27.

“There is a documented need for accessible, affordable, fresh food on college campuses,” says Heather Paxson, William R. Kenan, Jr. Professor of Anthropology and one of the organizers of the program. “The problems for a small farmer in finding a sufficient market … are connected to the challenges of food insecurity in even wealthy areas. And so, it really is about connecting those dots.”

Through the six weeks of the project, farm stand shoppers purchased more than 2,000 pounds of fresh produce that they wouldn’t otherwise have had access to. Hannan, Paxson, and the team hope that this year’s pilot was successful enough to continue into future growing seasons, either in this farm stand form or as something else that can equally serve the campus community.

“This year we decided to pour our heart, soul, and resources into this vision and prove what’s possible,” says Susy Jones, senior sustainability project manager at MITOS. “How can we do it in a way that is robust and goes through the official MIT channels, and yet pushes the boundaries of what’s possible at MIT?”

A growing idea

Mohammed Hannan, founder of Hannan Healthy Foods, first met Paxson and Jones in 2022. Jones was looking for someone local who grew vegetables common in Asian cuisine in response to a student request. Paxson wanted a small farm to host a field trip for her subject 21A.155 (Food, Culture and Politics). In July, Paxson and Jones learned about an article in the Boston Globe featuring Hannan as an example of a small farmer hit hard by federal budget cuts.

They knew right away they wanted to help. They pulled in Zachary Rapaport and Aleks Banas, architecture master’s students and the co-founders of MIT Farm, an organization dedicated to getting the MIT community off campus and onto local farms. This MIT contingent connected with Hannan to come up with a plan.

“These projects — when they flow, they flow,” says Jones. “There was so much common ground and excitement that we were all willing to jump on calls at 7 p.m. many nights to figure it out.”

After a series of rapid-fire brainstorming sessions, the group decided to host weekly volunteer sessions at Hannan’s farm during the autumn growing season and sell the harvest at a farm stand on campus.

“It fits in seamlessly with the MIT motto, ‘mind and hand,’ ‘mens et manus,’ learning by doing, as well as the heart, which has been added unofficially — mind, hand, heart,” says Paxson.

Jones tapped into the MITOS network for financial, operational, student, and city partners. Rapaport and Banas put out calls for volunteers. Paxson incorporated a volunteer trip into her syllabus and allocated discretionary project funding to subsidize the cost of the produce, allowing the food to be sold at 50 percent of the wholesale price that Hannan was paid for it.

“The fact that MIT students, faculty, and staff could come out to the farm, and that our harvest would circulate back to campus and into the broader community — there’s an energy around it that’s very different from academics. It feels essential to be part of something so tangible,” says Rapaport.

The volunteer sessions proved to be popular. Throughout the pilot, about 75 students and half a dozen faculty and staff trekked out to Lincoln from MIT’s Cambridge, Massachusetts, campus at least once to clear fields and harvest vegetables. Hannan hopes the experience will change the way they think about their food.

“Harvesting the produce, knowing the operation, knowing how hard it is, it’ll stick in their brain,” he says.

On that September Sunday, second-year electrical engineering and computer science major Abrianna Zhang had come out with a friend after seeing a notification on the dormspam email lists. Zhang grew up in a California suburb big on supporting local farmers, but volunteering showed her a different side of the job.

“There’s a lot of work that goes into raising all these crops and then getting all this manual labor,” says Zhang. “It makes me think about the economy of things. How is this even possible … for us to gain access to organic fruits or produce at a reasonable price?”

Setting up shop

Since mid-September, Monday has been Farm Stand day at MIT. Tables covered in green gingham tablecloths strike through the Stata Center lobby, holding stacks of cardboard boxes filled with produce. Customers wait in line to claim their piece of the fresh harvest — carrots, potatoes, onions, tomatoes, herbs, and various greens.

Many of these students typically head to off-campus grocery stores to get their fresh produce. Katie Stabb, a sophomore civil and environmental engineering major and self-proclaimed “crazy plant lady,” grows her own food in the summer, but travels far from campus to shop for her vegetables during the school year. Having this stand right at MIT gives her time back, and she’s been spreading the news to her East Campus dorm mates — even picking boxes up for them when they can’t make it themselves and helping them figure out what to do with their excess ingredients.

“I have encountered having way too many chives before, but that’s new for some folks,” she says. “Last week we pooled all of our chives and I made chive pancakes, kind of like scallion pancakes.”

Stabb is not alone. In a multi-question customer survey conducted at the close of the Farm Stand season, 62 percent of respondents said the Harvest Box gave them the chance to try new foods and 49 percent experimented with new recipes. Seventy percent said this project helped them increase their vegetable intake.

Nearly 60 percent of the survey respondents were graduate students living off campus. Banas, one of the MIT Farm co-leads, is one of those grad students enjoying the benefits.

“I was cooking and making food that I bought from the farm stand and thought, ‘Oh, this is very literally influencing my life in a positive way.’ And I’m hoping that this has a similar impact for other people,” she says.

The impact goes beyond the ability of students to nourish themselves with fresh vegetables. New communities have grown from this collaboration. Jones, for example, expanded her network at MITOS by tapping into expertise and resources from MIT Dining, the Vice President for Finance Merchant Services, and the MIT Federal Credit Union.

“There were just these pockets of people in every corner of MIT who know how to do these very specific things that might seem not very glamorous, but make something like this possible,” says Jones. “It’s such a positive, affirming moment when you’re starting from scratch and someone’s like, ‘This is such a cool idea, how can I help?’”

Strengthening community

Inviting people from MIT to connect across campus and explore beyond Cambridge has helped students and employees alike feel like they’re part of something bigger.

“The community that’s grown around this work is what keeps me so engaged,” says Rapaport. “MIT can have a bit of a siloing effect. It’s easy to become so focused on your classes and academics that your world revolves around them. Farm club grew out of wanting to build connections across the student body and to see ourselves and MIT as part of a larger network of people, communities, and relationships.”

This particular connection will continue to grow, as Rapaport and Banas will use their architectural expertise to lead a design-build team in developing a climate-adaptive and bio-based root cellar at Hannan Healthy Foods, to improve the farm’s winter vegetable storage conditions. 

Community engagement is an ethos Hannan has embraced since the start of his farming journey in 2018, motivated by a desire to provision first his family and then others with healthy food.

“One thing I have done over the years, I was not trying to do farming by myself,” he says. “I always reached out to as many people as I could. The idea is, if community is not involved, they just see it as an individual business.”

It’s why he gifts his volunteers huge bags of tomatoes at the end of a shift, or donates some of his harvest to food banks, or engages an advisory committee of local residents to ensure he’s filling the right needs.

“There’s a reciprocal dimension to gifting that needs to continue,” says Paxson. “That is what builds and maintains community — it’s classic anthropology."

And much of what’s exchanged in this type of reciprocity can’t be charted or graded or marked on a spreadsheet. It’s cooking pancakes with dorm mates. It’s meeting and appreciating new colleagues. It’s grabbing a friend to harvest cabbage on a beautiful autumn Sunday.

“Seeing a student who volunteered over the weekend harvesting chives come to the market on Monday and then want to take a selfie with those chives,” says Jones. “To me, that’s a cool moment.”

At a high level, ammonia seems like a dream fuel: It’s carbon-free, energy-dense, and easier to move and store than hydrogen. Ammonia is also already manufactured and transported at scale, meaning it could transform energy systems using existing infrastructure. But burning ammonia creates dangerous nitrous oxides, and splitting ammonia molecules to create hydrogen fuel typically requires lots of energy and specialized engines.

The startup Amogy, founded by four MIT alumni, believes it has the technology to finally unlock ammonia as a major fuel source. The company has developed a catalyst it says can split — or “crack” — ammonia into hydrogen and nitrogen up to 70 percent more efficiently than state-of-the-art systems today. The company is planning to sell its catalysts as well as modular systems including fuel cells and engines to convert ammonia directly to power. Those systems don’t burn or combust ammonia, and thus bypass the health concerns related to nitrous oxides.

Since Amogy’s founding in 2020, the company has used its ammonia-cracking technology to create the world’s first ammonia-powered drone, tractor, truck, and tugboat. It has also attracted partnerships with industry leaders including Samsung, Saudi Aramco, KBR, and Hyundai, raising more than $300 million along the way.

“No one has showcased that ammonia can be used to power things at the scale of ships and trucks like us,” says CEO Seonghoon Woo PhD ’15, who founded the company with Hyunho Kim PhD ’18, Jongwon Choi PhD ’17, and Young Suk Jo SM ’13, PhD ’16. “We’ve demonstrated this approach works and is scalable.”

Earlier this year, Amogy completed a research and manufacturing facility in Houston and announced a pilot deployment of its catalyst with the global engineering firm JGC Holdings Corporation. Now, with a manufacturing contract secured with Samsung Heavy Industries, Amogy is set to start delivering more of its systems to customers next year. The company will deploy a 1-megawatt ammonia-to-power pilot project with the South Korean city of Pohang in 2026, with plans to scale up to 40 megawatts at that site by 2028 or 2029. Woo says dozens of other projects with multinational corporations are in the works.

Because of the power density advantages of ammonia over renewables and batteries, the company is targeting power-hungry industries like maritime shipping, power generation, construction, and mining for its early systems.

“This is only the beginning,” Woo says. “We’ve worked hard to build the technology and the foundation of our company, but the real value will be generated as we scale. We’ve proved the potential for ammonia to decarbonize heavy industry, and now we really want to accelerate adoption of our technology. We’re thinking long term about the energy transition.”

Unlocking a new fuel source

Woo completed his PhD in MIT’s Department of Materials Science and Engineering before his eventual co-founders, Kim, Choi, and Jo, completed their PhDs in MIT’s Department of Mechanical Engineering. Jo worked on energy science and ran experiments to make engines run more efficiently as part of his PhD.

“The PhD programs at MIT teach you how to think deeply about solving technical problems using systems-based approaches,” Woo says. “You also realize the value in learning from failures, and that mindset of iteration is similar to what you need to do in startups.”

In 2020, Woo was working in the semiconductor industry when he reached out to his eventual co-founders asking if they were working on anything interesting. At that time, Jo was still working on energy systems based on hydrogen and ammonia while Kim was developing new catalysts to create ammonia fuel.

“I wanted to start a company and build a business to do good things for society,” Woo recalls. “People had been talking about hydrogen as a more sustainable fuel source, but it had never come to fruition. We thought there might be a way to improve ammonia catalyst technology and accelerate the hydrogen economy.”

The founders started experimenting with Jo’s technology for ammonia cracking, the process in which ammonia (NH3) molecules split into their nitrogen (N2) and hydrogen (H2) constituent parts. Ammonia cracking to date has been done at huge plants in high-temperature reactors that require large amounts of energy. Those high temperatures limited the catalyst materials that could be used to drive the reaction.

Starting from scratch, the founders were able to identify new material recipes that could be used to miniaturize the catalyst and work at lower temperatures. The proprietary catalyst materials allow the company to create a system that can be deployed in new places at lower costs.

“We really had to redevelop the whole technology, including the catalyst and reformer, and even the integration with the larger system,” Woo says. “One of the most important things is we don’t combust ammonia — we don’t need pilot fuel, and we don’t generate any nitrogen gas or CO2.”

Today Amogy has a portfolio of proprietary catalyst technologies that use base metals along with precious metals. The company has proven the efficiency of its catalysts in demonstrations beginning with the first ammonia-powered drone in 2021. The catalyst can be used to produce hydrogen more efficiently, and by integrating the catalyst with hydrogen fuel cells or engines, Amogy also offers modular ammonia-to-power systems that can scale to meet customer energy demands.

“We’re enabling the decarbonization of heavy industry,” Woo says. “We are targeting transportation, chemical production, manufacturing, and industries that are carbon-heavy and need to decarbonize soon, for example to achieve domestic goals. Our vision in the longer term is to enable ammonia as a fuel in a variety of applications, including power generation, first at microgrids and then eventually full grid-scale.”

Scaling with industry

When Amogy completed its facility in Houston, one of their early visitors was MIT Professor Evelyn Wang, who is also MIT’s vice president for energy and climate. Woo says other people involved in the Climate Project at MIT have been supportive.

Another key partner for Amogy is Samsung Heavy Industries, which announced a multiyear deal to manufacturing Amogy’s ammonia-to-power systems on Nov. 12.

“Our strategy is to partner with the existing big players in heavy industry to accelerate the commercialization of our technology,” Woo says. “We have worked with big oil and gas companies like BHP and Saudi Aramco, companies interested in hydrogen fuel like KBR and Mitsubishi, and many more industrial companies.”

When paired with other clean energy technologies to provide the power for its systems, Woo says Amogy offers a way to completely decarbonize sectors of the economy that can’t electrify on their own.

“In heavy transport, you have to use high-energy density liquid fuel because of the long distances and power requirements,” Woo says. “Batteries can’t meet those requirements. It’s why hydrogen is such an exciting molecule for heavy industry and shipping. But hydrogen needs to be kept super cold, whereas ammonia can be liquid at room temperature. Our job now is to provide that power at scale.”

There is growing attention on the links between artificial intelligence and increased energy demands. But while the power-hungry data centers being built to support AI could potentially stress electricity grids, increase customer prices and service interruptions, and generally slow the transition to clean energy, the use of artificial intelligence can also help the energy transition.

For example, use of AI is reducing energy consumption and associated emissions in buildings, transportation, and industrial processes. In addition, AI is helping to optimize the design and siting of new wind and solar installations and energy storage facilities.

On electric power grids, using AI algorithms to control operations is helping to increase efficiency and reduce costs, integrate the growing share of renewables, and even predict when key equipment needs servicing to prevent failure and possible blackouts. AI can help grid planners schedule investments in generation, energy storage, and other infrastructure that will be needed in the future. AI is also helping researchers discover or design novel materials for nuclear reactors, batteries, and electrolyzers.

Researchers at MIT and elsewhere are actively investigating aspects of those and other opportunities for AI to support the clean energy transition. At its 2025 research conference, MITEI announced the Data Center Power Forum, a targeted research effort for MITEI member companies interested in addressing the challenges of data center power demand.

Controlling real-time operations

Customers generally rely on receiving a continuous supply of electricity, and grid operators get help from AI to make that happen — while optimizing the storage and distribution of energy from renewable sources at the same time.

But with more installation of solar and wind farms — both of which provide power in smaller amounts, and intermittently — and the growing threat of weather events and cyberattacks, ensuring reliability is getting more complicated. “That’s exactly where AI can come into the picture,” explains Anuradha Annaswamy, a senior research scientist in MIT’s Department of Mechanical Engineering and director of MIT’s Active-Adaptive Control Laboratory. “Essentially, you need to introduce a whole information infrastructure to supplement and complement the physical infrastructure.”

The electricity grid is a complex system that requires meticulous control on time scales ranging from decades all the way down to microseconds. The challenge can be traced to the basic laws of power physics: electricity supply must equal electricity demand at every instant, or generation can be interrupted. In past decades, grid operators generally assumed that generation was fixed — they could count on how much electricity each large power plant would produce — while demand varied over time in a fairly predictable way. As a result, operators could commission specific power plants to run as needed to meet demand the next day. If some outages occurred, specially designated units would start up as needed to make up the shortfall.

Today and in the future, that matching of supply and demand must still happen, even as the number of small, intermittent sources of generation grows and weather disturbances and other threats to the grid increase. AI algorithms provide a means of achieving the complex management of information needed to forecast within just a few hours which plants should run while also ensuring that the frequency, voltage, and other characteristics of the incoming power are as required for the grid to operate properly.

Moreover, AI can make possible new ways of increasing supply or decreasing demand at times when supplies on the grid run short. As Annaswamy points out, the battery in your electric vehicle (EV), as well as the one charged up by solar panels or wind turbines, can — when needed — serve as a source of extra power to be fed into the grid. And given real-time price signals, EV owners can choose to shift charging from a time when demand is peaking and prices are high to a time when demand and therefore prices are both lower. In addition, new smart thermostats can be set to allow the indoor temperature to drop or rise —  a range defined by the customer — when demand on the grid is peaking. And data centers themselves can be a source of demand flexibility: selected AI calculations could be delayed as needed to smooth out peaks in demand. Thus, AI can provide many opportunities to fine-tune both supply and demand as needed.

In addition, AI makes possible “predictive maintenance.” Any downtime is costly for the company and threatens shortages for the customers served. AI algorithms can collect key performance data during normal operation and, when readings veer off from that normal, the system can alert operators that something might be going wrong, giving them a chance to intervene. That capability prevents equipment failures, reduces the need for routine inspections, increases worker productivity, and extends the lifetime of key equipment.

Annaswamy stresses that “figuring out how to architect this new power grid with these AI components will require many different experts to come together.” She notes that electrical engineers, computer scientists, and energy economists “will have to rub shoulders with enlightened regulators and policymakers to make sure that this is not just an academic exercise, but will actually get implemented. All the different stakeholders have to learn from each other. And you need guarantees that nothing is going to fail. You can’t have blackouts.”

Using AI to help plan investments in infrastructure for the future

Grid companies constantly need to plan for expanding generation, transmission, storage, and more, and getting all the necessary infrastructure built and operating may take many years, in some cases more than a decade. So, they need to predict what infrastructure they’ll need to ensure reliability in the future. “It’s complicated because you have to forecast over a decade ahead of time what to build and where to build it,” says Deepjyoti Deka, a research scientist in MITEI.

One challenge with anticipating what will be needed is predicting how the future system will operate. “That’s becoming increasingly difficult,” says Deka, because more renewables are coming online and displacing traditional generators. In the past, operators could rely on “spinning reserves,” that is, generating capacity that’s not currently in use but could come online in a matter of minutes to meet any shortfall on the system. The presence of so many intermittent generators — wind and solar — means there’s now less stability and inertia built into the grid. Adding to the complication is that those intermittent generators can be built by various vendors, and grid planners may not have access to the physics-based equations that govern the operation of each piece of equipment at sufficiently fine time scales. “So, you probably don’t know exactly how it’s going to run,” says Deka.

And then there’s the weather. Determining the reliability of a proposed future energy system requires knowing what it’ll be up against in terms of weather. The future grid has to be reliable not only in everyday weather, but also during low-probability but high-risk events such as hurricanes, floods, and wildfires, all of which are becoming more and more frequent, notes Deka. AI can help by predicting such events and even tracking changes in weather patterns due to climate change.

Deka points out another, less-obvious benefit of the speed of AI analysis. Any infrastructure development plan must be reviewed and approved, often by several regulatory and other bodies. Traditionally, an applicant would develop a plan, analyze its impacts, and submit the plan to one set of reviewers. After making any requested changes and repeating the analysis, the applicant would resubmit a revised version to the reviewers to see if the new version was acceptable. AI tools can speed up the required analysis so the process moves along more quickly. Planners can even reduce the number of times a proposal is rejected by using large language models to search regulatory publications and summarize what’s important for a proposed infrastructure installation.

Harnessing AI to discover and exploit advanced materials needed for the energy transition

“Use of AI for materials development is booming right now,” says Ju Li, MIT’s Carl Richard Soderberg Professor of Power Engineering. He notes two main directions.

First, AI makes possible faster physics-based simulations at the atomic scale. The result is a better atomic-level understanding of how composition, processing, structure, and chemical reactivity relate to the performance of materials. That understanding provides design rules to help guide the development and discovery of novel materials for energy generation, storage, and conversion needed for a sustainable future energy system.

And second, AI can help guide experiments in real time as they take place in the lab. Li explains: “AI assists us in choosing the best experiment to do based on our previous experiments and — based on literature searches — makes hypotheses and suggests new experiments.”

He describes what happens in his own lab. Human scientists interact with a large language model, which then makes suggestions about what specific experiments to do next. The human researcher accepts or modifies the suggestion, and a robotic arm responds by setting up and performing the next step in the experimental sequence, synthesizing the material, testing the performance, and taking images of samples when appropriate. Based on a mix of literature knowledge, human intuition, and previous experimental results, AI thus coordinates active learning that balances the goals of reducing uncertainty with improving performance. And, as Li points out, “AI has read many more books and papers than any human can, and is thus naturally more interdisciplinary.”

The outcome, says Li, is both better design of experiments and speeding up the “work flow.” Traditionally, the process of developing new materials has required synthesizing the precursors, making the material, testing its performance and characterizing the structure, making adjustments, and repeating the same series of steps. AI guidance speeds up that process, “helping us to design critical, cheap experiments that can give us the maximum amount of information feedback,” says Li.

“Having this capability certainly will accelerate material discovery, and this may be the thing that can really help us in the clean energy transition,” he concludes. “AI [has the potential to] lubricate the material-discovery and optimization process, perhaps shortening it from decades, as in the past, to just a few years.” 

MITEI’s contributions

At MIT, researchers are working on various aspects of the opportunities described above. In projects supported by MITEI, teams are using AI to better model and predict disruptions in plasma flows inside fusion reactors — a necessity in achieving practical fusion power generation. Other MITEI-supported teams are using AI-powered tools to interpret regulations, climate data, and infrastructure maps in order to achieve faster, more adaptive electric grid planning. AI-guided development of advanced materials continues, with one MITEI project using AI to optimize solar cells and thermoelectric materials.

Other MITEI researchers are developing robots that can learn maintenance tasks based on human feedback, including physical intervention and verbal instructions. The goal is to reduce costs, improve safety, and accelerate the deployment of the renewable energy infrastructure. And MITEI-funded work continues on ways to reduce the energy demand of data centers, from designing more efficient computer chips and computing algorithms to rethinking the architectural design of the buildings, for example, to increase airflow so as to reduce the need for air conditioning.

In addition to providing leadership and funding for many research projects, MITEI acts as a convenor, bringing together interested parties to consider common problems and potential solutions. In May 2025, MITEI’s annual spring symposium — titled “AI and energy: Peril and promise” — brought together AI and energy experts from across academia, industry, government, and nonprofit organizations to explore AI as both a problem and a potential solution for the clean energy transition. At the close of the symposium, William H. Green, director of MITEI and Hoyt C. Hottel Professor in the MIT Department of Chemical Engineering, noted, “The challenge of meeting data center energy demand and of unlocking the potential benefits of AI to the energy transition is now a research priority for MITEI.”

In April, MIT senior Josh Randolph will race 26.2 miles across Concord, Massachusetts, and neighboring towns, carrying a 50-lb backpack. The race, called the Tough Ruck, honors America’s fallen military and first responders. For Randolph, it is one of the most rewarding experiences he’s done in his time at MIT, and he’s never missed a race.

“I want to do things that are challenging and push me to learn more about myself,” says Randolph, a Nebraska native. “As soon as I found out about the Tough Ruck, I knew I was going to be a part of it.”

Carrying on tradition and honoring those before him is a priority for Randolph. Both of his grandfathers served in the United States Air Force, and now he’s following in their footsteps through leadership in the U.S. Air Force Reserve Officers’ Training Corps (AFROTC) at MIT. His work with MIT Emergency Medical Services (EMS) has inspired him to aim for medical school so he could join the Air Force as a doctor.

“I always wanted to be in public service, serve my community, and serve my country,” Randolph says.

Getting attached to medicine

Randolph was particularly close with his grandfather, who worked with electronics in the Air Force and later became an engineer.

“I’ve always seen him as a big role model of mine. He’s very proud of his service,” Randolph says. A mechanical engineering major, he shares his grandfather’s interest in the scientific and technical side of the military.

But Randolph hasn’t let his commitment to the Air Force narrow his experiences at MIT.

He signed up for MIT EMS in his sophomore year as a way to push out of his comfort zone. Although he didn’t have a strong interest in medicine at the time, he was excited about being responsible for providing essential services to his community.

“If somebody’s in need on campus, they call 911, and we’re entrusted with the responsibility to help them out and keep them safe. I didn’t even know that was something you could do in college,” Randolph says.

Getting late-night calls and handling high-pressure situations took some getting used to, but he loved that he was helping.

“It feels a little uncomfortable at first, but then the more calls you run, the more experience you get and the more comfortable you feel with it, and then the more you want to do,” Randolph says.

Since joining in his second year, Randolph has responded to more than 100 911 calls and now holds the rank of provincial crew chief, meaning he provides basic life support patient care and coordinates on-scene operations.

His experiences interacting with patients and racing around Cambridge, Massachusetts, to help his community made him realize he would regret not pursuing medicine. In his final year at MIT, he set his sights on medical school. “Even though it was pretty late, I decided to make that switch and put my all into medicine,” Randolph says.

After serving as class officer during his junior year, helping to oversee the EMT certification process, Randolph became the director of professional development in his senior year. In this role, he oversees the training and development of service members as well as the quality of patient care. “It’s great to see how new students integrate and gain bigger roles and become more involved with the services,” Randolph says. “It’s really rewarding to contribute a little bit toward their development within EMS and then also just as people.”

Leadership in the ROTC

Randolph knew he would be a part of Air Force ROTC since early in high school. He later earned the Air Force ROTC Type 1 scholarship that gave him a tuition-free spot at MIT. It was through AFROTC that he became further committed to helping and honoring those around him, including through the Tough Ruck.

“Pretty often there are family members of fallen servicemembers who make tags with their loved one’s name on it and they hand them out for people to carry on their rucks, which is pretty cool, Randolph said of the race. “Overall, it is a really supportive environment, and I try to give as many people high fives and as much encouragement as I can, but at some point I get too tired and need to focus on running.”

His parents come out to watch every year.

In previous semesters, Randolph has served as flight commander and group commander within AFROTC’s Detachment 365, which is based at MIT and also hosts cadets from Harvard University, Tufts University, and Wellesley College. Currently, as squadron commander, he leads one of the 20-cadet units that makes up the detachment. He has co-organized three Leadership Laboratories dedicated to training over 70 cadets.

Randolph has earned the AFROTC Field Training Superior Performance Award, the AFROTC Commendation Award, the AFROTC Achievement Award, and the Military Order of the World Wars Bronze Award. He has also received the AFROTC Academic Honors Award five times, the Physical Fitness Award four times, and the Maximum AFROTC Physical Fitness Assessment Award two times. 

He keeps his activities and schoolwork straight through to-do lists and calendar items, but he admits the workload can still be tough.

“One thing that has helped me is trying to prioritize and figure out what things need my attention immediately or what things will be very important. If it is something that is important and will affect or benefit a lot of people, I try and devote my energy toward that to make the most of my time and implement meaningful things,” Randolph says.

A human-centered direction

For the last two years, Randolph worked in the Pappalardo Laboratory as an apprentice and undergraduate assistant, helping students design, fabricate, and test robots they were building for a class design challenge. He has also conducted linguistics research with Professor Suzanne Flynn and worked in the labs of professor of nuclear science and engineering Michael Short and professor of biological and mechanical engineering Domatilla Del Vecchio.

Randolph has also volunteered his time through English for Speakers of Other Languages, where he worked as a volunteer to help MIT employees improve their English speaking and writing skills.

For now, he is excited to enter a more human-centered field through his studies in medicine. After watching his father survive two bouts of cancer, thanks in part to robotically assisted surgery, he hopes to develop robotic health care applications.

“I want to have a deeper and more tangible connection to people. Compassion and empathy are things that I really want to try and live by,” Randolph says. “I think being the most empathetic and compassionate with the people you take care of is always a good thing.”

Brian Hanna, operations manager of MIT Venture Mentoring Service (VMS), connects skilled volunteer mentors with MIT entrepreneurs looking to launch, expand, and enhance their vision.  

MIT VMS is a free service, supporting innovation across the Institute, available to all current MIT students, staff members, faculty members, or alums of a degree-granting program living in the Greater Boston area. If a community member has an idea that they’d like help developing, Hanna and his team will match them with a team of mentors who can provide practical, as-needed expertise and knowledge to guide your venture. 

VMS is part of the MIT ecosystem for entrepreneurs. VMS mentors are selected for their experience in areas relevant to entrepreneurs’ needs and assist with a range of business challenges, including marketing, finance, and product development. As the program celebrates its 25th anniversary of serving MIT’s entrepreneurial community, it has supported more than 3,500 ventures and mentored over 4,800 participants. 

When Hanna began working at VMS in 2023, he was new to the program but not to the Institute. Prior to joining VMS, he served as the employer relations coordinator in Career Advising and Professional Development (CAPD), where he worked with companies interested in recruiting MIT talent. His responsibilities included organizing career fairs, scheduling interviews, and building relationships with various local employers. After two years at CAPD, Hanna transitioned to the role of center coordinator at the McGovern Institute for Brain Research. While Hanna does not claim to be a neuroscientist, his organizational skills proved valuable as he supported six different research centers at McGovern, with research ranging from autism to bionics.  

As the VMS operations manager, Hanna supervises staff members who run events and boot camps and schedule an average of 50 mentoring sessions a month. Whether it’s a first-time entrepreneur who comes up with an idea on their morning commute or an industry veteran with licensing and a patent in place, Hanna strategically matches them with mentors who can help them build their skill set and grow their business. Hanna also provides oversight to over 200 volunteer VMS mentors, half of which are MIT alumni. 

In addition to processing all incoming applications (about 25 per month), Hanna also oversees a monthly mentor meeting centered around strengthening the VMS mentor community. During the meeting, the VMS team shares announcements, discusses upcoming events, hosts guest speakers, and invites a group of current ventures to give four-minute pitches for additional advice. These pitches allow mentees to receive input from the entire mentor network, rather than just their mentor team.   

The relationship between mentees, mentors, and VMS does not have an expiration date. Hanna notes that a saying in the office is, “we are VMS for life.” This rings true, as some ventures and mentors have been a part of the program for most of its 25-year existence. 

When a mentee is ready to meet with their mentors for the first time, VMS aims to schedule an in-person meeting to create a strong relationship. After that, the program embraces the flexibility of meeting via Zoom to help make scheduling easier. One of the most valuable resources outside of the mentoring sessions is the theme-specific boot camps sprinkled throughout the year. These sessions are four- or five-hour events led by mentors who cover topics such as marketing, business-to-business sales, or building an IP portfolio. They serve as crash courses where mentees can learn the basics of important aspects of entrepreneurship. Another resource offered to active mentees is office hours with experts in areas such as human resources, legal, and accounting. 

In December, VMS will celebrate its 25th anniversary with an event honoring current and former mentors. The event will look back on 25 years of impact and look ahead to the future of the program. 

Soundbytes 

Q: Do you have an MIT memory or project that brings you pride? 

Hanna: At the McGovern Institute, I was part of a team that worked on the first board meeting and launch event for the K. Lisa Yang Center for Bionics, which was an incredible experience. It was a brand-new research center led by world-class researchers and innovators. Since it was the first board meeting it was a big deal, so we planned to host a celebration tied to the meeting. There were a lot of moving parts and collaboration between faculty, researchers, staff, board members, and vendors. It took place at the tail end of Covid, which was an added challenge. With such an important event you don’t want to let anyone down. In the end, it worked out really well, was a fun event to be a part of, and something I never thought I would be able to do.

Q: How would you describe the community at MIT? 

Hanna: Very welcoming. I was intimidated when I first interviewed at MIT because, as someone who isn’t a STEM person, MIT was never on my radar. Then a job came up, and I thought, I'll apply for that. When I started working here, there was always someone available to provide assistance and point me in the right direction. Everyone is incredibly talented and innovative — not just in creating things, but also in problem-solving and finding ways to collaborate. Each time I changed roles, everyone I met was down-to-earth, kind, and extremely helpful during onboarding. It was never sink or swim — it was always nurturing. 

Q: What advice would you give to a new staff member at MIT? 

Hanna: Make connections with people outside of your immediate network. Get involved in the community by attending events or reaching out to people. For both jobs which I held after working at CAPD, I reached out to the hiring manager when I saw the job posting and asked a couple clarifying questions. Also, it’s important to know that everything is numbered; the buildings, the majors, everything.  

Some 200 light years from Earth, the core of a dead star is circling a larger star in a macabre cosmic dance. The dead star is a type of white dwarf that exerts a powerful magnetic field as it pulls material from the larger star into a swirling, accreting disk. The spiraling pair is what’s known as an “intermediate polar” — a type of star system that gives off a complex pattern of intense radiation, including X-rays, as gas from the larger star falls onto the other one.

Now, MIT astronomers have used an X-ray telescope in space to identify key features in the system’s innermost region — an extremely energetic environment that has been inaccessible to most telescopes until now. In an open-access study published in the Astrophysical Journal, the team reports using NASA’s Imaging X-ray Polarimetry Explorer (IXPE) to observe the intermediate polar, known as EX Hydrae.

The team found a surprisingly high degree of X-ray polarization, which describes the direction of an X-ray wave’s electric field, as well as an unexpected direction of polarization in the X-rays coming from EX Hydrae. From these measurements, the researchers traced the X-rays back to their source in the system’s innermost region, close to the surface of the white dwarf.

What’s more, they determined that the system’s X-rays were emitted from a column of white-hot material that the white dwarf was pulling in from its companion star. They estimate that this column is about 2,000 miles high — about half the radius of the white dwarf itself and much taller than what physicists had predicted for such a system. They also determined that the X-rays are reflected off the white dwarf’s surface before scattering into space — an effect that physicists suspected but hadn’t confirmed until now.

The team’s results demonstrate that X-ray polarimetry can be an effective way to study extreme stellar environments such as the most energetic regions of an accreting white dwarf.

“We showed that X-ray polarimetry can be used to make detailed measurements of the white dwarf's accretion geometry,” says Sean Gunderson, a postdoc in MIT’s Kavli Institute for Astrophysics and Space Research, who is the study’s lead author. “It opens the window into the possibility of making similar measurements of other types of accreting white dwarfs that also have never had predicted X-ray polarization signals.”

Gunderson’s MIT Kavli co-authors include graduate student Swati Ravi and research scientists Herman Marshall and David Huenemoerder, along with Dustin Swarm of the University of Iowa, Richard Ignace of East Tennessee State University, Yael Nazé of the University of Liège, and Pragati Pradhan of Embry Riddle Aeronautical University.

A high-energy fountain

All forms of light, including X-rays, are influenced by electric and magnetic fields. Light travels in waves that wiggle, or oscillate, at right angles to the direction in which the light is traveling. External electric and magnetic fields can pull these oscillations in random directions. But when light interacts and bounces off a surface, it can become polarized, meaning that its vibrations tighten up in one direction. Polarized light, then, can be a way for scientists to trace the source of the light and discern some details about the source’s geometry.

The IXPE space observatory is NASA’s first mission designed to study polarized X-rays that are emitted by extreme astrophysical objects. The spacecraft, which launched in 2021, orbits the Earth and records these polarized X-rays. Since launch, it has primarily focused on supernovae, black holes, and neutron stars.

The new MIT study is the first to use IXPE to measure polarized X-rays from an intermediate polar — a smaller system compared to black holes and supernovas, that nevertheless is known to be a strong emitter of X-rays.

“We started talking about how much polarization would be useful to get an idea of what’s happening in these types of systems, which most telescopes see as just a dot in their field of view,” Marshall says.

An intermediate polar gets its name from the strength of the central white dwarf’s magnetic field. When this field is strong, the material from the companion star is directly pulled toward the white dwarf’s magnetic poles. When the field is very weak, the stellar material instead swirls around the dwarf in an accretion disk that eventually deposits matter directly onto the dwarf’s surface.

In the case of an intermediate polar, physicists predict that material should fall in a complex sort of in-between pattern, forming an accretion disk that also gets pulled toward the white dwarf’s poles. The magnetic field should lift the disk of incoming material far upward, like a high-energy fountain, before the stellar debris falls toward the white dwarf’s magnetic poles, at speeds of millions of miles per hour, in what astronomers refer to as an “accretion curtain.” Physicists suspect that this falling material should run up against previously lifted material that is still falling toward the poles, creating a sort of traffic jam of gas. This pile-up of matter forms a column of colliding gas that is tens of millions of degrees Fahrenheit and should emit high-energy X-rays.

An innermost picture

By measuring any polarized X-rays emitted by EX Hydrae, the team aimed to test the picture of intermediate polars that physicists had hypothesized. In January 2025, IXPE took a total of about 600,000 seconds, or about seven days’ worth, of X-ray measurements from the system.

“With every X-ray that comes in from the source, you can measure the polarization direction,” Marshall explains. “You collect a lot of these, and they’re all at different angles and directions which you can average to get a preferred degree and direction of the polarization.”

Their measurements revealed an 8 percent polarization degree that was much higher than what scientists had predicted according to some theoretical models. From there, the researchers were able to confirm that the X-rays were indeed coming from the system’s column, and that this column is about 2,000 miles high.

“If you were able to stand somewhat close to the white dwarf’s pole, you would see a column of gas stretching 2,000 miles into the sky, and then fanning outward,” Gunderson says.

The team also measured the direction of EX Hydrae’s X-ray polarization, which they determined to be perpendicular to the white dwarf’s column of incoming gas. This was a sign that the X-rays emitted by the column were then bouncing off the white dwarf’s surface before traveling into space, and eventually into IXPE’s telescopes.

“The thing that’s helpful about X-ray polarization is that it’s giving you a picture of the innermost, most energetic portion of this entire system,” Ravi says. “When we look through other telescopes, we don’t see any of this detail.”

The team plans to apply X-ray polarization to study other accreting white dwarf systems, which could help scientists get a grasp on much larger cosmic phenomena.

“There comes a point where so much material is falling onto the white dwarf from a companion star that the white dwarf can’t hold it anymore, the whole thing collapses and produces a type of supernova that’s observable throughout the universe, which can be used to figure out the size of the universe,” Marshall offers. “So understanding these white dwarf systems helps scientists understand the sources of those supernovae, and tells you about the ecology of the galaxy.”

This research was supported, in part, by NASA.

Large language models (LLMs) like ChatGPT can write an essay or plan a menu almost instantly. But until recently, it was also easy to stump them. The models, which rely on language patterns to respond to users’ queries, often failed at math problems and were not good at complex reasoning. Suddenly, however, they’ve gotten a lot better at these things.

A new generation of LLMs known as reasoning models are being trained to solve complex problems. Like humans, they need some time to think through problems like these — and remarkably, scientists at MIT’s McGovern Institute for Brain Research have found that the kinds of problems that require the most processing from reasoning models are the very same problems that people need take their time with. In other words, they report today in the journal PNAS, the “cost of thinking” for a reasoning model is similar to the cost of thinking for a human.

The researchers, who were led by Evelina Fedorenko, an associate professor of brain and cognitive sciences and an investigator at the McGovern Institute, conclude that in at least one important way, reasoning models have a human-like approach to thinking. That, they note, is not by design. “People who build these models don’t care if they do it like humans. They just want a system that will robustly perform under all sorts of conditions and produce correct responses,” Fedorenko says. “The fact that there’s some convergence is really quite striking.”

Reasoning models

Like many forms of artificial intelligence, the new reasoning models are artificial neural networks: computational tools that learn how to process information when they are given data and a problem to solve. Artificial neural networks have been very successful at many of the tasks that the brain’s own neural networks do well — and in some cases, neuroscientists have discovered that those that perform best do share certain aspects of information processing in the brain. Still, some scientists argued that artificial intelligence was not ready to take on more sophisticated aspects of human intelligence.

“Up until recently, I was among the people saying, ‘These models are really good at things like perception and language, but it’s still going to be a long ways off until we have neural network models that can do reasoning,” Fedorenko says. “Then these large reasoning models emerged and they seem to do much better at a lot of these thinking tasks, like solving math problems and writing pieces of computer code.”

Andrea Gregor de Varda, a K. Lisa Yang ICoN Center Fellow and a postdoc in Fedorenko’s lab, explains that reasoning models work out problems step by step. “At some point, people realized that models needed to have more space to perform the actual computations that are needed to solve complex problems,” he says. “The performance started becoming way, way stronger if you let the models break down the problems into parts.”

To encourage models to work through complex problems in steps that lead to correct solutions, engineers can use reinforcement learning. During their training, the models are rewarded for correct answers and penalized for wrong ones. “The models explore the problem space themselves,” de Varda says. “The actions that lead to positive rewards are reinforced, so that they produce correct solutions more often.”

Models trained in this way are much more likely than their predecessors to arrive at the same answers a human would when they are given a reasoning task. Their stepwise problem-solving does mean reasoning models can take a bit longer to find an answer than the LLMs that came before — but since they’re getting right answers where the previous models would have failed, their responses are worth the wait.

The models’ need to take some time to work through complex problems already hints at a parallel to human thinking: if you demand that a person solve a hard problem instantaneously, they’d probably fail, too. De Varda wanted to examine this relationship more systematically. So he gave reasoning models and human volunteers the same set of problems, and tracked not just whether they got the answers right, but also how much time or effort it took them to get there.

Time versus tokens

This meant measuring how long it took people to respond to each question, down to the millisecond. For the models, Varda used a different metric. It didn’t make sense to measure processing time, since this is more dependent on computer hardware than the effort the model puts into solving a problem. So instead, he tracked tokens, which are part of a model’s internal chain of thought. “They produce tokens that are not meant for the user to see and work on, but just to have some track of the internal computation that they’re doing,” de Varda explains. “It’s as if they were talking to themselves.”

Both humans and reasoning models were asked to solve seven different types of problems, like numeric arithmetic and intuitive reasoning. For each problem class, they were given many problems. The harder a given problem was, the longer it took people to solve it — and the longer it took people to solve a problem, the more tokens a reasoning model generated as it came to its own solution.

Likewise, the classes of problems that humans took longest to solve were the same classes of problems that required the most tokens for the models: arithmetic problems were the least demanding, whereas a group of problems called the “ARC challenge,” where pairs of colored grids represent a transformation that must be inferred and then applied to a new object, were the most costly for both people and models.

De Varda and Fedorenko say the striking match in the costs of thinking demonstrates one way in which reasoning models are thinking like humans. That doesn’t mean the models are recreating human intelligence, though. The researchers still want to know whether the models use similar representations of information to the human brain, and how those representations are transformed into solutions to problems. They’re also curious whether the models will be able to handle problems that require world knowledge that is not spelled out in the texts that are used for model training.

The researchers point out that even though reasoning models generate internal monologues as they solve problems, they are not necessarily using language to think. “If you look at the output that these models produce while reasoning, it often contains errors or some nonsensical bits, even if the model ultimately arrives at a correct answer. So the actual internal computations likely take place in an abstract, non-linguistic representation space, similar to how humans don’t use language to think,” he says.

As an undergraduate majoring in architecture, Dong Nyung Lee ’21 wasn’t sure how to respond when friends asked him what the study of architecture was about.

“I was always confused about how to describe it myself,” he says with a laugh. “I would tell them that it wasn’t just about a building, or a city, or a community. It’s a balance across different scales, and it has to touch everything all at once.”

As a graduate student enrolled in a design studio course last spring — 4.154 (Territory as Interior) — Lee and his classmates had to design a building that would serve a specific community in a specific location. The course, says Lee, gave him clarity as to “what architecture is all about.”

Designed by Roi Salgueiro Barrio, a lecturer in the MIT School of Architecture and Planning’s Department of Architecture, the coursework combines ecological principles, architectural design, urban economics, and social considerations to address real-world problems in marginalized or degraded areas.

“When we build, we always impact economies, mostly by the different types of technologies we use and their dependence on different types of labor and materials,” says Salgueiro Barrio. “The intention here was to think at both levels: the activities that can be accommodated, and how we can actually build something.”

Research first

Students were tasked with repurposing an abandoned fishing industry building on the Barbanza Peninsula in Galicia, Spain, and proposing a new economic activity for the building that would help regenerate the local economy. Working in groups, they researched the region’s material resources and fiscal sectors and designed detailed maps. This approach to constructing a building was new for Vincent Jackow a master's student in architecture.

“Normally in architecture, we work at the scale of one-to-100 meters,” he says. But this process allowed me to connect the dots between what the region offered and what could be built to support the economy.”

The aim of revitalizing this area is also a goal of Fundación RIA (FRIA), a nonprofit think tank established by Pritzker Prize-winning architect David Chipperfield. FRIA generates research and territorial planning with the goal of long-term sustainability of the built and natural environment in the Galicia region. During their spring break in March, the students traveled to Galicia, met with Chipperfield, business owners, fishermen, and farmers, and explored a variety of sites. They also consulted with the owner of the building they were to repurpose.

Returning to MIT, the students constructed nine detailed models. Master’s student Aleks Banaś says she took the studio because it required her to explore the variety of scales in an architectural project from territorial analysis to building detail, all while keeping the socio-economic aspect of design decisions in mind.

“I’m interested in how architecture can support local economies,” says Banaś. “Visiting Galicia was very special because of the communities we interacted with. We were no longer looking at articles and maps of the region; we were learning about day-to-day life. A lot of people shared with us the value of their work, which is not economically feasible.”

Banaś was impressed by the region’s strong maritime history and the generations of craftspeople working on timber boat-making. Inspired by the collective spirit of the region, she designed “House of Sea,” transforming the former cannery into a hub for community gathering and seafront activities. The reimagined building would accommodate a variety of functions including a boat-building workshop for the Ribeira carpenters’ association, a restaurant, and a large, covered section for local events such as the annual barnacle festival.

“I wanted to demonstrate how we can create space for an alternative economy that can host and support these skills and traditions,” says Banaś. 

Jackow’s building — “La Nueva Cordelería,” or “New Rope Making” — was a facility using hemp to produce rope and hempcrete blocks (a construction material). The production of both “is very on-trend in the E.U.” and provides an alternative to petrochemical-based ropes for the region’s marine uses, says Jackow. The building would serve as a cultural hub, incorporating a café, worker housing, and offices. Even its very structure would also make use of the rope by joining timber with knots allowing the interior spaces to be redesigned.

Lee’s building was designed to engage with the forestry and agricultural industries.

“What intrigued me was that Galicia is heavily dependent on pulp production and wood harvesting,” he says. “I wanted to give value to the post-harvest residue.”

Lee designed a biochar plant using some of the concrete and terra cotta blocks on site. Biochar is made by heating the harvested wood residue through pyrolysis — thermal decomposition in an environment with little oxygen. The resulting biochar would be used by farmers for soil enhancement.

“The work demonstrated an understanding of the local resources and using them to benefit the revitalization of the area,” says Salgueiro Barrio, who was pleased with the results. 

FRIA was so impressed with the work that they held an exhibition at their gallery in Santiago de Compostela in August and September to highlight the importance of connecting academic research with the territory through student projects. Banaś interned with FRIA over the summer working on multiple projects, including the plan and design for the exhibition. The challenge here, she says, was to design an exhibition of academic work for a general audience. The final presentation included maps, drawings, and photographs by the students.

For Lee, the course was more meaningful than any he has taken to date. Moving between the different scales of the project illustrated, for him, “the biggest challenge for a designer and an architect. Architecture is universal, and very specific. Keeping those dualities in focus was the biggest challenge and the most interesting part of this project. It hit at the core of what architecture is.”

Taking an audience of hundreds on a tour around the body, seven speakers at The Picower Institute for Learning and Memory’s symposium “Circuits of Survival and Homeostasis” Oct. 21 shared their advanced and novel research about some of the nervous system’s most evolutionarily ancient functions.

Introducing the symposium that she arranged with a picture of a man at a campfire on a frigid day, Sara Prescott, assistant professor in the Picower Institute and MIT’s departments of Biology and Brain and Cognitive Sciences, pointed out that the brain and the body cooperate constantly just to keep us going, and that when the systems they maintain fail, the consequence is disease.

“[This man] is tightly regulating his blood pressure, glucose levels, his energy expenditure, inflammation and breathing rate, and he’s doing this in the face of a fluctuating external environment,” Prescott said. “Behind each of these processes there are networks of neurons that are working quietly in the background to maintain internal stability. And this is, of course, the brain’s oldest job.”

Indeed, although the discoveries they shared about the underlying neuroscience were new, the speakers each described experiences that are as timeless as they are familiar: the beating of the heart, the transition from hunger to satiety, and the healing of cuts on our skin.

Feeling warm and full

Li Ye, a scientist at Scripps Research, picked right up on the example of coping with the cold. Mammals need to maintain a consistent internal body temperature, and so they will increase metabolism in the cold and then, as energy supplies dwindle, seek out more food. His lab’s 2023 study identified the circuit, centered in the Xiphoid nucleus of the brain’s thalamus, that regulates this behavior by sensing prolonged cold exposure and energy consumption. Ye described other feeding mechanisms his lab is studying as well, including searching out the circuitry that regulates how long an animal will feed at a time. For instance, if you’re worried about predators finding you, it’s a bad idea to linger for a leisurely lunch.

Physiologist Zachary Knight of the University of California at San Francisco also studies feeding and drinking behaviors. In particular, his lab asks how the brain knows when it’s time to stop. The conventional wisdom is that all that’s needed is a feeling of fullness coming from the gut, but his research shows there is more to the story. A 2023 study from his lab found a population of neurons in the caudal nucleus of the solitary tract in the brain stem that receive signals about ingestion and taste from the mouth, and that send that “stop eating” signal. They also found a separate neural population in the brain stem that indeed receives fullness signals from the gut, and teaches the brain over time how much food leads to satisfaction. Both neuron types work together to regulate the pace of eating. His lab has continued to study how brain stem circuits regulate feeding using these multiple inputs.

Energy balance depends not only on how many calories come in, but also on how much energy is spent. When food is truly scarce, many animals will engage in a state of radically lowered metabolism called torpor (like hibernation), where body temperature plummets. The brain circuits that exert control over body temperature are another area of active research. In his talk, Harvard University neurologist Clifford Saper described years of research in which his lab found neurons in the median preoptic nucleus that dictate this metabolic state. Recently, his lab demonstrated that the same neurons that regulate torpor also regulate fever during sickness. When the neurons are active, body temperature drops. When they are inhibited, fever ensues. Thus, the same neurons act as a two-way switch for body temperature in response to different threatening conditions.

Sickness, injury, and stress

As the idea of fever suggests, the body also has evolved circuits (that scientists are only now dissecting) to deal with sickness and injury.

Washington University neuroscientist Qin Liu described her research into the circuits governing coughing and sneezing, which, on one hand, can clear the upper airways of pathogens and obstructions but, on the other hand, can spread those pathogens to others in the community. She described her lab’s 2024 study in which her team pinpointed a population of neurons in the nasal passages that mediate sneezing and a different population of sensory neurons in the trachea that produce coughing. Identifying the specific cells and their unique characteristics makes them potentially viable drug targets.

While Liu tackled sickness, Harvard stem cell biologist Ya-Chieh Hsu discussed how neurons can reshape the body’s tissues during stress and injury, specifically the hair and skin. While it is common lore that stress can make your hair gray and fall out, Hsu’s lab has shown the actual physiological mechanisms that make it so. In 2020 her team showed that bursts of noradrenaline from the hyperactivation of nerves in the sympathetic nervous system kills the melanocyte stem cells that give hair its color. She described newer research indicating a similar mechanism may also make hair fall out by killing off cells at the base of hair follicles, releasing cellular debris and triggering auto-immunity. Her lab has also looked at how the nervous system influences skin healing after injury. For instance, while our skin may appear to heal after a cut because it closes up, many skin cell types actually don’t rebound (unless you’re still an embryo). By looking at the difference between embryos and post-birth mice, Hsu’s lab has traced the neural mechanisms that prevent fuller healing, identifying a role for cells called fibroblasts and the nervous system.

Continuing on the theme of stress, Caltech biologist Yuki Oka discussed a broad-scale project in his lab to develop a molecular and cellular atlas of the sympathetic nervous system, which innervates much of the body and famously produces its “fight or flight” responses. In work partly published last year, their journey touched on cells and circuits involved in functions ranging from salivation to secreting bile. Oka and co-authors made the case for the need to study the system more in a review paper earlier this year.

A new model to study human biology

In their search for the best ways to understand the circuits that govern survival and homeostasis, researchers often use rodents because they are genetically tractable, easy to house, and reproduce quickly, but Stanford University biochemist Mark Krasnow has worked to develop a new model with many of those same traits but a closer genetic relationship to humans: the mouse lemur. In his talk, he described that work (which includes extensive field research in Madagascar) and focused on insights the mouse lemurs have helped him make into heart arrhythmias. After studying the genes and health of hundreds of mouse lemurs, his lab identified a family with “sick sinus syndrome,” an arrhythmia also seen in humans. In a preprint study, his lab describes the specific molecular pathways at fault in disrupting the heart’s natural pace making.

By sharing some of the latest research into how the brain and body work to stay healthy, the symposium’s speakers highlighted the most current thinking about the nervous system’s most primal purposes.

Ernest Opoku knew he wanted to become a scientist when he was a little boy. But his school in Dadease, a small town in Ghana, offered no elective science courses — so Opoku created one for himself.

Even though they had neither a dedicated science classroom nor a lab, Opoku convinced his principal to bring in someone to teach him and five other friends he had convinced to join him. With just a chalkboard and some imagination, they learned about chemical interactions through the formulas and diagrams they drew together.

“I grew up in a town where it was difficult to find a scientist,” he says.

Today, Opoku has become one himself, recently earning a PhD in quantum chemistry from Auburn University. This year, he joins MIT as a part of the School of Science Dean’s Postdoctoral Fellowship program. Working with the Van Voorhis Group at the Department of Chemistry, Opoku’s goal is to advance computational methods to study how electrons behave — a fundamental research that underlies applications ranging from materials science to drug discovery.

“As a boy who wanted to satisfy my own curiosities at a young age, in addition to the fact that my parents had minimal formal education,” Opoku says, “I knew that the only way I would be able to accomplish my goal was to work hard.”

In pursuit of knowledge

When Opoku was 8 years old, he began independently learning English at school. He would come back with homework, but his parents were unable to help him, as neither of them could read or write in English. Frustrated, his mother asked an older student to help tutor her son.

Every day, the boys would meet at 6 o’clock. With no electricity at either of their homes, they practiced new vocabulary and pronunciations together by a kerosene lamp.

As he entered junior high school, Opoku’s fascination with nature grew.

“I realized that chemistry was the central science that really offered the insight that I wanted to really understand Creation from the smallest level,” he says.

He studied diligently and was able to get into one of Ghana’s top high schools — but his parents couldn’t afford the tuition. He therefore enrolled in Dadease Agric Senior High School in his hometown. By growing tomatoes and maize, he saved up enough money to support his education.

In 2012, he got into Kwame Nkrumah University of Science and Technology (KNUST), a first-ranking university in Ghana and the West Africa region. There, he was introduced to computational chemistry. Unlike many other branches of science, the field required only a laptop and the internet to study chemical reactions.

“Anything that comes to mind, anytime I can grab my computer and I’ll start exploring my curiosity. I don’t have to wait to go to the laboratory in order to interrogate nature,” he says.

Opoku worked from early morning to late night. None of it felt like work, though, thanks to his supervisor, the late quantum chemist Richard Tia, who was an associate professor of chemistry at KNUST.

“Every single day was a fun day,” he recalls of his time working with Tia. “I was being asked to do the things that I myself wanted to know, to satisfy my own curiosity, and by doing that I’ll be given a degree.”

In 2020, Opoku’s curiosity brought him even further, this time overseas to Auburn University in Alabama for his PhD. Under the guidance of his advisor, Professor J. V. Ortiz, Opoku contributed to the development of new computational methods to simulate how electrons bind to or detach from molecules, a process known as electron propagation.

What is new about Opoku’s approach is that it does not rely on any adjustable or empirical parameters. Unlike some earlier computational methods that require tuning to match experimental results, his technique uses advanced mathematical formulations to directly account for first principles of electron interactions. This makes the method more accurate — closely resembling results from lab experiments — while using less computational power.

By streamlining the calculations and eliminating guesswork, Opoku’s work marks a major step toward faster, more trustworthy quantum simulations across a wide range of molecules, including those never studied before — laying the groundwork for breakthroughs in many areas such as materials science and sustainable energy.

For his postdoctoral research at MIT, Opoku aims to advance electron propagator methods to address larger and more complex molecules and materials by integrating quantum computing, machine learning, and bootstrap embedding — a technique that simplifies quantum chemistry calculations by dividing large molecules into smaller, overlapping fragments. He is collaborating with Troy Van Voorhis, the Haslam and Dewey Professor of Chemistry, whose expertise in these areas can help make Opoku’s advanced simulations more computationally efficient and scalable.

“His approach is different from any of the ways that we've pursued in the group in the past,” Van Voorhis says.

Passing along the opportunity to learn

Opoku thanks previous mentors who helped him overcome the “intellectual overhead required to make contributions to the field,” and believes Van Voorhis will offer the same kind of support.

In 2021, Opoku joined the National Organization for the Professional Advancement of Black Chemists and Chemical Engineers (NOBCChE) to gain mentorship, networking, and career development opportunities within a supportive community. He later led the Auburn University chapter as president, helping coordinate k-12 outreach to inspire the next generation of scientists, engineers, and innovators.

“Opoku’s mentorship goes above and beyond what would be typical at his career stage,” says Van Voorhis. “One reason is his ability to communicate science to people, and not just the concepts of science, but also the process of science."

Back home, Opoku founded the Nesvard Institute of Molecular Sciences to support African students to develop not only skills for graduate school and professional careers, but also a sense of confidence and cultural identity. Through the nonprofit, he has mentored 29 students so far, passing along the opportunity for them to follow their curiosity and help others do the same.

“There are many areas of science and engineering to which Africans have made significant contributions, but these contributions are often not recognized, celebrated, or documented,” Opoku says.

He adds: “We have a duty to change the narrative.” 

Computer-Aided Design (CAD) is the go-to method for designing most of today’s physical products. Engineers use CAD to turn 2D sketches into 3D models that they can then test and refine before sending a final version to a production line. But the software is notoriously complicated to learn, with thousands of commands to choose from. To be truly proficient in the software takes a huge amount of time and practice.

MIT engineers are looking to ease CAD’s learning curve with an AI model that uses CAD software much like a human would. Given a 2D sketch of an object, the model quickly creates a 3D version by clicking buttons and file options, similar to how an engineer would use the software.

The MIT team has created a new dataset called VideoCAD, which contains more than 41,000 examples of how 3D models are built in CAD software. By learning from these videos, which illustrate how different shapes and objects are constructed step-by-step, the new AI system can now operate CAD software much like a human user.

With VideoCAD, the team is building toward an AI-enabled “CAD co-pilot.” They envision that such a tool could not only create 3D versions of a design, but also work with a human user to suggest next steps, or automatically carry out build sequences that would otherwise be tedious and time-consuming to manually click through.

“There’s an opportunity for AI to increase engineers’ productivity as well as make CAD more accessible to more people,” says Ghadi Nehme, a graduate student in MIT’s Department of Mechanical Engineering.

“This is significant because it lowers the barrier to entry for design, helping people without years of CAD training to create 3D models more easily and tap into their creativity,” adds Faez Ahmed, associate professor of mechanical engineering at MIT.

Ahmed and Nehme, along with graduate student Brandon Man and postdoc Ferdous Alam, will present their work at the Conference on Neural Information Processing Systems (NeurIPS) in December.

Click by click

The team’s new work expands on recent developments in AI-driven user interface (UI) agents — tools that are trained to use software programs to carry out tasks, such as automatically gathering information online and organizing it in an Excel spreadsheet. Ahmed’s group wondered whether such UI agents could be designed to use CAD, which encompasses many more features and functions, and involves far more complicated tasks than the average UI agent can handle.

In their new work, the team aimed to design an AI-driven UI agent that takes the reins of the CAD program to create a 3D version of a 2D sketch, click by click. To do so, the team first looked to an existing dataset of objects that were designed in CAD by humans. Each object in the dataset includes the sequence of high-level design commands, such as “sketch line,” “circle,” and “extrude,” that were used to build the final object.

However, the team realized that these high-level commands alone were not enough to train an AI agent to actually use CAD software. A real agent must also understand the details behind each action. For instance: Which sketch region should it select? When should it zoom in? And what part of a sketch should it extrude? To bridge this gap, the researchers developed a system to translate high-level commands into user-interface interactions.

“For example, let’s say we drew a sketch by drawing a line from point 1 to point 2,” Nehme says. “We translated those high-level actions to user-interface actions, meaning we say, go from this pixel location, click, and then move to a second pixel location, and click, while having the ‘line’ operation selected.”

In the end, the team generated over 41,000 videos of human-designed CAD objects, each of which is described in real-time in terms of the specific clicks, mouse-drags, and other keyboard actions that the human originally carried out. They then fed all this data into a model they developed to learn connections between UI actions and CAD object generation.

Once trained on this dataset, which they dub VideoCAD, the new AI model could take a 2D sketch as input and directly control the CAD software, clicking, dragging, and selecting tools to construct the full 3D shape. The objects ranged in complexity from simple brackets to more complicated house designs. The team is training the model on more complex shapes and envisions that both the model and the dataset could one day enable CAD co-pilots for designers in a wide range of fields.

“VideoCAD is a valuable first step toward AI assistants that help onboard new users and automate the repetitive modeling work that follows familiar patterns,” says Mehdi Ataei, who was not involved in the study, and is a senior research scientist at Autodesk Research, which develops new design software tools. “This is an early foundation, and I would be excited to see successors that span multiple CAD systems, richer operations like assemblies and constraints, and more realistic, messy human workflows.”

If there was one thing Cameron Halliday SM ’19, MBA ’22, PhD ’22 was exceptional at during the early days of his PhD at MIT, it was producing the same graph over and over again. Unfortunately for Halliday, the graph measured various materials’ ability to absorb CO2 at high temperatures over time — and it always pointed down and to the right. That meant the materials lost their ability to capture the molecules responsible for warming our climate.

At least Halliday wasn’t alone: For many years, researchers have tried and mostly failed to find materials that could reliably absorb CO2 at the super-high temperatures of industrial furnaces, kilns, and boilers. Halliday’s goal was to find something that lasted a little longer.

Then in 2019, he put a type of molten salt called lithium-sodium ortho-borate through his tests. The salts absorbed more than 95 percent of the CO2. And for the first time, the graph showed almost no degradation over 50 cycles.  The same was true after 100 cycles. Then 1,000.

“I honestly don’t know if we ever expected to completely solve the problem,” Halliday says. “We just expected to improve the system. It took another two months to figure out why it worked.”

The researchers discovered the salts behave like a liquid at high temperatures, which avoids the brittle cracking responsible for the degradation of many solid materials.

“I remember walking home over the Mass Ave bridge at 5 a.m. with all the morning runners going by me,” Halliday recalls. “That was the moment when I realized what this meant. Since then, it’s been about proving it works at larger scales. We’ve just been building the next scaled-up version, proving it still works, building a bigger version, proving that out, until we reach the ultimate goal of deploying this everywhere.”

Today, Halliday is the co-founder and CEO of Mantel, a company building systems to capture carbon dioxide at large industrial sites of all types. Although a lot of people think the carbon capture industry is a dead end, Halliday doesn’t give up so easily, and he’s got a growing corpus of performance data to keep him encouraged.

Mantel’s system can be added on to the machines of power stations and factories making cement, steel, paper and pulp, oil and gas, and more, reducing their carbon emissions by around 95 percent. Instead of being released into the atmosphere, the emitted CO2 is channeled into Mantel’s system, where the company’s salts are sprayed out from something that looks like a shower head. The CO2 diffuses through the molten salts in a reaction that can be reversed through further temperature increases, so the salts boil off pure CO2 that can be transported for use or stored underground.

A key difference from other carbon capture methods that have struggled to be profitable is that Mantel uses the heat from its process to generate steam for customers by combining it with water in another part of its system. Mantel says delivering steam, which is used to drive many common industrial processes, lets its system work with just 3 percent of the net energy that state-of-the-art carbon capture systems require.

“We’re still consuming energy, but we get most of it back as steam, whereas the incumbent technology only consumes steam,” says Halliday, who co-founded Mantel with Sean Robertson PhD ’22 and Danielle Rapson. “That steam is a useful revenue stream, so we can turn carbon capture from a waste management process into a value creation process for our customer’s core business — whether that’s a power station using steam to make electricity, or oil and gas refineries. It completely changes the economics of carbon capture.”

From science to startup

Halliday’s first exposure to MIT came in 2016 when he cold emailed Alan Hatton, MIT’s Ralph Landau Professor of Chemical Engineering Practice, asking if he could come to his lab for the summer and work on research into carbon capture.

“He invited me, but he didn’t put me on that project,” Halliday recalls. “At the end of the summer he said, ‘You should consider coming back and doing a PhD.’”

Halliday enrolled in a joint PhD-MBA program the following year.

“I really wanted to work on something that had an impact,” Halliday says. “The dual PhD-MBA program has some deep technical academic elements to it, but you also work with a company for two months, so you use a lot of what you learn in the real world.”

Halliday worked on a few different research projects in Hatton’s lab early on, all three of which eventually turned into companies. The one that he stuck with explored ways to make carbon capture more energy efficient by working at the high temperatures common at emissions-heavy industrial sites.

Halliday ran into the same problems as past researchers with materials degrading at such extreme conditions.

“It was the big limiter for the technology,” Halliday recalls.

Then Halliday ran his successful experiment with molten borate salts in 2019. The MBA portion of his program began soon after, and Halliday decided to use that time to commercialize the technology. Part of that occurred in Course 15.366 (Climate and Energy Ventures), where Halliday met his co-founders. As it happens, alumni of the class have started more than 150 companies over the years.

“MIT tries to pull these great ideas out of academia and get them into the world so they can be valued and used,” Halliday says. “For the Climate and Energy Ventures class, outside speakers showed us every stage of company-building. The technology roadmap for our system is shoebox-sized, shipping container, one-bedroom house, and then the size of a building. It was really valuable to see other companies and say, ‘That’s what we could look like in three years, or six years.”

From startup to scale up

When Mantel was officially founded in 2022 the founders had their shoebox-sized system. After raising early funding, the team built its shipping container-sized system at The Engine, an MIT-affiliated startup incubator. That system has been operational for almost two years.

Last year, Mantel announced a partnership with Kruger Inc. to build the next version of its system at a factory in Quebec, which will be operational next year. The plant will run in a two-year test phase before scaling across Kruger’s other plants if successful.

“The Quebec project is proving the capture efficiency and proving the step-change improvement in energy use of our system,” Halliday says. “It’s a derisking of the technology that will unlock a lot more opportunities.”

Halliday says Mantel is in conversations with close to 100 industrial partners around the world, including the owners of refineries, data centers, cement and steel plants, and oil and gas companies. Because it’s a standalone addition, Halliday says Mantel’s system doesn’t have to change much to be used in different industries.

Mantel doesn’t handle CO2 conversion or sequestration, but Halliday says capture makes up the bulk of the costs in the CO2 value chain. It also generates high-quality CO2 that can be transported in pipelines and used in industries including the food and beverage industry — like the CO2 that makes your soda bubbly.

“This is the solution our customers are dreaming of,” Halliday says. “It means they don’t have to shut down their billion-dollar asset and reimagine their business to address an issue that they all appreciate is existential. There are questions about the timeline, but most industries recognize this is a problem they’ll have to grapple with eventually. This is a pragmatic solution that’s not trying to reshape the world as we dream of it. It’s looking at the problem at hand today and fixing it.”

Anchorage-dependent cells are cells that require physical attachment to a solid surface, such as a culture dish, to survive, grow, and reproduce. In the biomedical industry, and others, having the ability to culture these cells is crucial, but current techniques used to separate cells from surfaces can induce stresses and reduce cell viability.

“In the pharmaceutical and biotechnology industries, cells are typically detached from culture surfaces using enzymes — a process fraught with challenges,” says Kripa Varanasi, MIT professor of mechanical engineering. “Enzymatic treatments can damage delicate cell membranes and surface proteins, particularly in primary cells, and often require multiple steps that make the workflow slow and labor-intensive.”

Existing approaches also rely on large volumes of consumables, generating an estimated 300 million liters of cell culture waste each year. Moreover, because these enzymes are often animal-derived, they can introduce compatibility concerns for cells intended for human therapies, limiting scalability and high-throughput applications in modern biomanufacturing.

Varanasi is corresponding author on a new paper in the journal ACS Nano, in which researchers from the MIT Department of Mechanical Engineering and the Cancer Program at the Broad Institute of Harvard and MIT present a novel enzyme-free strategy for detaching cells from culture surfaces. The method works by harnessing alternating electrochemical current on a conductive biocompatible polymer nanocomposite surface.

“By applying low-frequency alternating voltage, our platform disrupts adhesion within minutes while maintaining over 90 percent cell viability — overcoming the limitations of enzymatic and mechanical methods that can damage cells or generate excess waste,” says Varanasi.

Beyond simplifying routine cell culture, the approach could transform large-scale biomanufacturing by enabling automated and contamination-conscious workflows for cell therapies, tissue engineering, and regenerative medicine. The platform also provides a pathway for safely expanding and harvesting sensitive immune cells for applications such as CAR-T therapies.

“Because our electrically tunable interface can dynamically shape the ionic microenvironment around cells, it also offers powerful opportunities to control ion channels, study signaling pathways, and integrate with bioelectronic systems for high-throughput drug screening, regenerative medicine, and personalized therapies,” Varanasi explains.

“Our work shows how electrochemistry can be harnessed not just for scientific discovery, but also for scalable, real-world applications,” says Wang Hee (Wren) Lee, MIT postdoc and co-first author. “By translating electrochemical control into biomanufacturing, we’re laying the foundation for technologies that can accelerate automation, reduce waste, and ultimately enable new industries built on sustainable and precise processing.”

Bert Vandereydt, co-first author and mechanical engineering researcher at MIT, emphasizes the potential for industrial scalability. “Because this method can be applied uniformly across large areas, it’s ideal for high-throughput and large-scale applications like cell therapy manufacturing. We envision it enabling fully automated, closed-loop cell culture systems in the near future.”

Yuen-Yi (Moony) Tseng, principal investigator at the Broad Institute and collaborator on the project, underscores the biomedical significance. “This platform opens new doors for culturing and harvesting delicate primary or cancer cells. It could streamline workflows across research and clinical biomanufacturing, reducing variability and preserving cell functionality for therapeutic use.”

Industrial applications of adherent cells include uses in the biomedical, pharmaceutical, and cosmetic sectors. For this study, the team tested their new method using human cancer cells, including osteosarcoma and ovarian cancer cells. After identifying an optimal frequency, the detachment efficiency for both types of cells increased from 1 percent to 95 percent, with cell viability exceeding 90 percent.

The paper, “Alternating Electrochemical Redox-Cycling on Nanocomposite Biointerface for High-Efficiency Enzyme-Free Cell Detachment,” is available from the American Chemical Society journal ACS Nano. 

“We’re here to talk about really substantive changes, and we want you to be a participant in that,” said Desirée Plata, the School of Engineering Distinguished Professor of Climate and Energy in MIT’s Department of Civil and Environmental Engineering, at Energizing@MIT: the MIT Energy Initiative’s (MITEI) Annual Research Conference that was held on Sept. 9-10.

Plata’s words resonated with the 150-plus participants from academia, industry, and government meeting in Cambridge for the conference, whose theme was “tackling emerging energy challenges.” Meeting such challenges and ultimately altering the trajectory of global climate outcomes requires partnerships, speakers agreed.

“We have to be humble and open,” said Giacomo Silvestri, chair of Eniverse Ventures at Eni, in a shared keynote address. “We cannot develop innovation just focusing on ourselves and our competencies … so we need to partner with startups, venture funds, universities like MIT and other public and private institutions.” 

Added his Eni colleague, Annalisa Muccioli, head of research and technology, “The energy transition is a race we can win only by combining mature solutions ready to deploy, together with emerging technologies that still require acceleration and risk management.”

Research targets

In a conference that showcased a suite of research priorities MITEI has identified as central to ensuring a low-carbon energy future, participants shared both promising discoveries and strategies for advancing proven technologies in the face of shifting political winds and policy uncertainties.

One panel focused on grid resiliency — a topic that has moved from the periphery to the center of energy discourse as climate-driven disruptions, cyber threats, and the integration of renewables challenge legacy systems. A dramatic case in point: the April 2025 outage in Spain and Portugal that left millions without power for eight to 15 hours. 

“I want to emphasize that this failure was about more than the power system,” said MITEI research scientist Pablo Duenas-Martinez. While he pinpointed technical problems with reactive power and voltage control behind the system collapse, Duenas-Martinez also called out a lack of transmission capacity with Central Europe and out-of-date operating procedures, and recommended better preparation and communication among transmission systems and utility operators.

“You can’t plan for every single eventuality, which means we need to broaden the portfolio of extreme events we prepare for,” noted Jennifer Pearce, vice president at energy company Avangrid. “We are making the system smarter, stronger, and more resilient to better protect from a wide range of threats such as storms, flooding, and extreme heat events.” Pearce noted that Avangrid’s commitment to deliver safe, reliable power to its customers necessitates “meticulous emergency planning procedures.”

The resiliency of the electric grid under greatly increased demand is an important motivation behind MITEI’s September 2025 launch of the Data Center Power Forum, which was also announced during the annual research conference. The forum will include research projects, webinars, and other content focused on energy supply and storage, grid design and management, infrastructure, and public and economic policy related to data centers. The forum’s members include MITEI companies that also participate in MIT’s Center for Environmental and Energy Policy Research (CEEPR).

Storage and transportation: Staggering challenges

Meeting climate goals to decarbonize the world by 2050 requires building around 300 terawatt-hours of storage, according to Asegun Henry, a professor in the MIT Department of Mechanical Engineering. “It’s an unbelievably enormous problem people have to wrap their minds around,” he said. Henry has been developing a high-temperature thermal energy storage system he has nicknamed “sun in a box.” His system uses liquid metal and graphite to hold electricity as heat and then convert it back to electricity, enabling storage anywhere from five to 500 hours.

“At the end of the day, storage provides a service, and the type of technology that you need is a function of the service that you value the most,” said Nestor Sepulveda, commercial lead for advanced energy investments and partnerships at Google. “I don't think there is one winner-takes-all type of market here.”

Another panel explored sustainable fuels that could help decarbonize hard-to-electrify sectors like aviation, shipping, and long-haul trucking. Randall Field, MITEI’s director of research, noted that sustainably produced drop-in fuels — fuels that are largely compatible with existing engines — “could eliminate potentially trillions of dollars of cost for fleet replacement and for infrastructure build-out, while also helping us to accelerate the rate of decarbonization of the transportation sectors."

Erik G. Birkerts is the chief growth officer of LanzaJet, which produces a drop-in, high-energy-density aviation fuel derived from agricultural residue and other waste carbon sources. “The key to driving broad sustainable aviation fuel adoption is solving both the supply-side challenge through more production and the demand-side hurdle by reducing costs,” he said.

“We think a good policy framework [for sustainable fuels] would be something that is technology-neutral, does not exclude any pathways to produce, is based on life cycle accounting practices, and on market mechanisms,” said Veronica L. Robertson, energy products technology portfolio manager at ExxonMobil.

MITEI plans a major expansion of its research on sustainable fuels, announcing a two-year study, “The future of fuels: Pathways to sustainable transportation,” starting in early 2026. According to Field, the study will analyze and assess biofuels and e-fuels.

Solutions from labs big and small

Global energy leaders offered glimpses of their research projects. A panel on carbon capture in power generation featured three takes on the topic: Devin Shaw, commercial director of decarbonization technologies at Shell, described post-combustion carbon capture in power plants using steam for heat recovery; Jan Marsh, a global program lead at Siemens Energy, discussed deploying novel materials to capture carbon dioxide directly from the air; and Jeffrey Goldmeer, senior director of technology strategy at GE Vernova, explained integrating carbon capture into gas-powered turbine systems.

During a panel on vehicle electrification, Brian Storey, vice president of energy and materials at the Toyota Research Institute, provided an overview of Toyota’s portfolio of projects for decarbonization, including solid-state batteries, flexible manufacturing lines, and grid-forming inverters to support EV charging infrastructure.

A session on MITEI seed fund projects revealed promising early-stage research inside MIT’s own labs. A new process for decarbonizing the production of ethylene was presented by Yogesh Surendranath, Donner Professor of Science in the MIT Department of Chemistry. Materials Science and Engineering assistant professor Aristide Gumyusenge also discussed the development of polymers essential for a new kind of sodium-ion battery.

Shepherding bold, new technologies like these from academic labs into the real world cannot succeed without ample support and deft management. A panel on paths to commercialization featured the work of Iwnetim Abate, Chipman Career Development Professor and assistant professor in the MIT Department of Materials Science and Engineering, who has spun out a company, Addis Energy, based on a novel geothermal process for harvesting clean hydrogen and ammonia from subsurface, iron-rich rocks. Among his funders: ARPA-E and MIT’s own The Engine Ventures.

The panel also highlighted the MIT Proto Ventures Program, an initiative to seize early-stage MIT ideas and unleash them as world-changing startups. “A mere 4.2 percent of all the patents that are actually prosecuted in the world are ever commercialized, which seems like a shocking number,” said Andrew Inglis, an entrepreneur working with Proto Ventures to translate geothermal discoveries into businesses. “Can’t we do this better? Let’s do this better!”

Geopolitical hazards

Throughout the conference, participants often voiced concern about the impacts of competition between the United States and China. Kelly Sims Gallagher, dean of the Fletcher School at Tufts University and an expert on China’s energy landscape, delivered the sobering news in her keynote address: “U.S. competitiveness in low-carbon technologies has eroded in nearly every category,” she said. “The Chinese are winning the clean tech race.”

China enjoys a 51 percent share in global wind turbine manufacture and 75 percent in solar modules. It also controls low-carbon supply chains that much of the world depends on. “China is getting so dominant that nobody can carve out a comparative advantage in anything,” said Gallagher. “China is just so big, and the scale is so huge that the Chinese can truly conquer markets and make it very hard for potential competitors to find a way in.”

And for the United States, the problem is “the seesaw of energy policy,” she says. “It’s incredibly difficult for the private sector to plan and to operate, given the lack of predictability and policy here.”

Nevertheless, Gallagher believes the United States still has a chance of at least regaining competitiveness, by setting up a stable, bipartisan energy policy, rebuilding domestic manufacturing and supply chains; providing consistent fiscal incentives; attracting and retaining global talent; and fostering international collaboration.

The conference shone a light on one such collaboration: a China-U.S. joint venture to manufacture lithium iron phosphate batteries for commercial vehicles in the United States. The venture brings together Eve Energy, a Chinese battery technology and manufacturing company; Daimler, a global commercial vehicle manufacturer; PACCAR Inc., a U.S.-based truck manufacturer; and Accelera, the zero-emissions business of Cummins Inc. “Manufacturing batteries in the U.S. makes the supply chain more robust and reduces geopolitical risks,” said Mike Gerty, of PACCAR.

While she acknowledged the obstacles confronting her colleagues in the room, Plata nevertheless concluded her remarks as a panel moderator with some optimism: “I hope you all leave this conference and look back on it in the future, saying I was in the room when they actually solved some of the challenges standing between now and the future that we all wish to manifest.”

MIT and GE Vernova launched the MIT-GE Vernova Energy and Climate Alliance on Sept. 15, a collaboration to advance research and education focused on accelerating the global energy transition.

Through the alliance — an industry-academia initiative conceived by MIT Provost Anantha Chandrakasan and GE Vernova CEO Scott Strazik — GE Vernova has committed $50 million over five years in the form of sponsored research projects and philanthropic funding for research, graduate student fellowships, internships, and experiential learning, as well as professional development programs for GE Vernova leaders.

“MIT has a long history of impactful collaborations with industry, and the collaboration between MIT and GE Vernova is a shining example of that legacy,” said Chandrakasan in opening remarks at a launch event. “Together, we are working on energy and climate solutions through interdisciplinary research and diverse perspectives, while providing MIT students the benefit of real-world insights from an industry leader positioned to bring those ideas into the world at scale.”

The energy of change

An independent company since its spinoff from GE in April 2024, GE Vernova is focused on accelerating the global energy transition. The company generates approximately 25 percent of the world’s electricity — with the world’s largest installed base of over 7,000 gas turbines, about 57,000 wind turbines, and leading-edge electrification technology.

GE Vernova’s slogan, “The Energy of Change,” is reflected in decisions such as locating its headquarters in Cambridge, Massachusetts — in close proximity to MIT. In pursuing transformative approaches to the energy transition, the company has identified MIT as a key collaborator.

A key component of the mission to electrify and decarbonize the world is collaboration, according to CEO Scott Strazik. “We want to inspire, and be inspired by, students as we work together on our generation’s greatest challenge, climate change. We have great ambition for what we want the world to become, but we need collaborators. And we need folks that want to iterate with us on what the world should be from here.”

Representing the Healey-Driscoll administration at the launch event were Massachusetts Secretary of Energy and Environmental Affairs Rebecca Tepper and Secretary of the Executive Office of Economic Development Eric Paley. Secretary Tepper highlighted the Mass Leads Act, a $1 billion climate tech and life sciences initiative enacted by Governor Maura Healey last November to strengthen Massachusetts’ leadership in climate tech and AI.

“We're harnessing every part of the state, from hydropower manufacturing facilities to the blue-to-blue economy in our south coast, and right here at the center of our colleges and universities. We want to invent and scale the solutions to climate change in our own backyard,” said Tepper. “That’s been the Massachusetts way for decades.”

Real-world problems, insights, and solutions

The launch celebration featured interactive science displays and student presenters introducing the first round of 13 research projects led by MIT faculty. These projects focus on generating scalable solutions to our most pressing challenges in the areas of electrification, decarbonization, renewables acceleration, and digital solutions. Read more about the funded projects here.

Collaborating with industry offers the opportunity for researchers and students to address real-world problems informed by practical insights. The diverse, interdisciplinary perspectives from both industry and academia will significantly strengthen the research supported through the GE Vernova Fellowships announced at the launch event.

“I’m excited to talk to the industry experts at GE Vernova about the problems that they work on,” said GE Vernova Fellow Aaron Langham. “I’m looking forward to learning more about how real people and industries use electrical power.”

Fellow Julia Estrin echoed a similar sentiment: “I see this as a chance to connect fundamental research with practical applications — using insights from industry to shape innovative solutions in the lab that can have a meaningful impact at scale.”

GE Vernova’s commitment to research is also providing support and inspiration for fellows. “This level of substantive enthusiasm for new ideas and technology is what comes from a company that not only looks toward the future, but also has the resources and determination to innovate impactfully,” says Owen Mylotte, a GE Vernova Fellow.

The inaugural cohort of eight fellows will continue their research at MIT with tuition support from GE Vernova. Find the full list of fellows and their research topics here.

Pipeline of future energy leaders

Highlighting the alliance’s emphasis on cultivating student talent and leadership, GE Vernova CEO Scott Strazik introduced four MIT alumni who are now leaders at GE Vernova: Dhanush Mariappan SM ’03, PhD ’19, senior engineering manager in the GE Vernova Advanced Research Center; Brent Brunell SM ’00, technology director in the Advanced Research Center; Paolo Marone MBA ’21, CFO of wind; and Grace Caza MAP ’22, chief of staff in supply chain and operations.

The four shared their experiences of working with MIT as students and their hopes for the future of this alliance in the realm of “people development,” as Mariappan highlighted. “Energy transition means leaders. And every one of the innovative research and professional education programs that will come out of this alliance is going to produce the leaders of the energy transition industry.”

The alliance is underscoring its commitment to developing future energy leaders by supporting the New Engineering Education Transformation program (NEET) and expanding opportunities for student internships. With 100 new internships for MIT students announced in the days following the launch, GE Vernova is opening broad opportunities for MIT students at all levels to contribute to a sustainable future.

“GE Vernova has been a tremendous collaborator every step of the way, with a clear vision of the technical breakthroughs we need to affect change at scale and a deep respect for MIT’s strengths and culture, as well as a hunger to listen and learn from us as well,” said Betar Gallant, alliance director who is also the Kendall Rohsenow Associate Professor of Mechanical Engineering at MIT. “Students, take this opportunity to learn, connect, and appreciate how much you’re valued, and how bright your futures are in this area of decarbonizing our energy systems. Your ideas and insight are going to help us determine and drive what’s next.”

Daring to create the future we want

The launch event transformed MIT’s Lobby 13 with green lighting and animated conversation around the posters and hardware demos on display, reflecting the sense of optimism for the future and the type of change the alliance — and the Commonwealth of Massachusetts — seeks to advance.

“Because of this collaboration and the commitment to the work that needs doing, many things will be created,” said Secretary Paley. “People in this room will work together on all kinds of projects that will do incredible things for our economy, for our innovation, for our country, and for our climate.”

The alliance builds on MIT’s growing portfolio of initiatives around sustainable energy systems, including the Climate Project at MIT, a presidential initiative focused on developing solutions to some of the toughest barriers to an effective global climate response. “This new alliance is a significant opportunity to move the needle of energy and climate research as we dare to create the future that we want, with the promise of impactful solutions for the world,” said Evelyn Wang, MIT vice president for energy and climate, who attended the launch.

To that end, the alliance is supporting critical cross-institution efforts in energy and climate policy, including funding three master’s students in MIT Technology and Policy Program and hosting an annual symposium in February 2026 to advance interdisciplinary research. GE Vernova is also providing philanthropic support to the MIT Human Insight Collaborative. For 2025-26, this support will contribute to addressing global energy poverty by supporting the MIT Abdul Latif Jameel Poverty Action Lab (J-PAL) in its work to expand access to affordable electricity in South Africa.

“Our hope to our fellows, our hope to our students is this: While the stakes are high and the urgency has never been higher, the impact that you are going to have over the decades to come has never been greater,” said Roger Martella, chief corporate and sustainability officer at GE Vernova. “You have so much opportunity to move the world in a better direction. We need you to succeed. And our mission is to serve you and enable your success.”

With the alliance’s launch — and GE Vernova’s new membership in several other MIT consortium programs related to sustainability, automation and robotics, and AI, including the Initiative for New Manufacturing, MIT Energy Initiative, MIT Climate and Sustainability Consortium, and Center for Transportation and Logistics — it’s evident why Betar Gallant says the company is “all-in at MIT.”

The potential for tremendous impact on the energy industry is clear to those involved in the alliance. As GE Vernova Fellow Jack Morris said at the launch, “This is the beginning of something big.”