As millions of acres of Canadian forest burned in the summer of 2025, the second-worst fire season in the country’s history, smoke drifted south, settling in the Upper Midwest, prompting air quality alerts from local regulatory agencies. For several hours on the last day of July, Chicago had the dirtiest air in the world, a scumble of smoke dimming the tall-towered skyline.

During these hazy days, David Keith, professor of geophysical sciences at the University of Chicago, was completing the move into his new office on the ground floor of Ryerson Hall. It is a high-ceilinged space with pale wood floors and old doors large enough to ride a horse through. Keith had been recruited to the University from Harvard just over two years earlier, in April of 2023, and charged with establishing the Climate Systems Engineering Initiative.

Keith is tall, narrow, bearded, bespectacled; a loping walker; and a warm and engaged conversationalist. When he alights upon a topic of acute interest, he hunches in his chair and begins to expound while gazing at the floor.

He is a controversial figure in the field of climate science—and in the broader world of environmental policy and activism—because he has devoted nearly all of his academic career to studying methods by which humans might deliberately intervene in the climate system to mitigate the effects of global warming. Much of this work has focused on adding sulfur to the stratosphere to reduce the amount of sunlight that reaches Earth—a skywide umbrella of aerosols to cool us just a bit, just enough.

Keith came to this work sideways. He studied physics as an undergraduate at the University of Toronto. After graduation he worked odd jobs, lived in a 12-person co-op, and spent time rock climbing in New York’s Shawangunk Mountains. He then went on to a doctorate in experimental physics at MIT where, under the guidance of David Pritchard, he helped build one of the world’s first atom interferometers. The work was “a thrill,” Keith has written, but, in some ways, hollow. If it sated his intellectual appetite, it did little else. He was troubled, too, that the Office of Naval Research funded the work: One of the interferometer’s immediate applications was improving the locational accuracy of submarines carrying ballistic missiles.

Keith was doing this work in the late ’80s, as global environmental catastrophe gained a public audience. The British Antarctic Survey reported on a large hole in Earth’s ozone layer. NASA scientist James Hansen testified before the Senate about the warming effects of greenhouse gas emissions. These were topics that Keith engaged with, vigorously, in an informal group of students from Boston-area colleges who met to discuss environmental science and policy.

The topics enthralled him, drawing on a deep personal connection to the natural world. His father was a field biologist researching toxic chemicals for the Canadian Wildlife Service. As a teenager Keith was a member of the Macoun Club, the youth arm of the Ottawa Field Naturalists’ Club, alongside a handful of bookish biology students who liked to camp. He hiked the Appalachian Trail solo from Maine to Vermont as a 17-year-old, and shortly after skied alone across Ontario’s Algonquin Provincial Park. Climate change, as he began to learn about it, inspired complex questions in precisely the way atomic interferometry did not. It raised policy implications of the largest magnitude; it was rich with intriguing and contentious scientific uncertainties.

Through a family connection, Keith met and talked with Hadi Dowlatabadi, a professor who was studying climate change in Carnegie Mellon University’s Department of Engineering and Public Policy. It was a productive chat. In 1991 Keith moved to Pittsburgh, leaving interferometry behind to begin a postdoc with Dowlatabadi.

Keith published his first paper on geoengineering one year later. The general idea had been around for decades. A 1965 report to President Lyndon B. Johnson—which, notably, spends a good deal of time fretting over human-induced climate change—discusses the potential of blanketing oceans in reflective particles. Soviet and American scientists in the ’70s wrote about creating a “stratospheric smog” to deflect sunlight. The idea surfaces and sinks, again and again, in academic literature, in government reports.

Geoengineering is a catchall term for a set of technologies and processes that range from planting trees to launching a quilt of light-deflecting mirrors the size of Brazil high into orbit between Earth and the sun. But the subject, in its diversity, can be broken into two broad approaches: those intended to pull carbon dioxide from the air, and those that cool Earth but do nothing about carbon.

Among the many ways to deflect the sun—formally known as solar radiation modification, or SRM—injecting sulfur into the stratosphere is the approach that’s likely best understood. Atmospheric dynamics in the stratosphere have been studied ever since scientists in the 1950s, eager to monitor nuclear fallout, used high-elevation U-2 flights to measure the concentration of radionuclides. Volcanic eruptions, meanwhile, have afforded centuries of insight into the cooling effects of sulfates in the stratosphere. Researchers have theorized about SRM aerosols that are potentially less pernicious than sulfates, which are known to deplete the ozone and cause pollution at ground level, but sulfates are “the devil we know,” Keith says.

This is not to suggest we know all we should. Keith and his collaborators readily admit to the need for more rigorous assessment by a more diverse group of researchers. “It’s really shocking that we don’t have a federal research program moving aggressively to sort out the physics, the chemistry, and the coupling dynamics [of this technology] in the context of climate,” says Jim Anderson, a climate scientist at Harvard who advised Keith during his second postdoc. “It’s unacceptable. We have to attack this problem with an entirely new level of innovation and aggression.”

But over the years, through his writing and speaking, Keith has outlined the broad shape of a potential SRM program. It would use hydrogen sulfide and begin with custom-designed aircraft that would be able to fly to roughly 20 kilometers, twice the altitude of commercial aircraft. “There’s nothing that’s unobtanium about it,” Keith says. Spy planes peered from these heights during the Cold War. But to do so while carrying a roughly one-ton payload presents a novel engineering challenge. At altitude, the planes would release the hydrogen sulfide, descend, reload, fly again. The amount deposited would start out small, on the order of one hundred thousand tons annually; assuming no major problems, this figure would increase each year up to a defined plateau.

Once let loose in the stratosphere, the hydrogen sulfide molecules would undergo a process of oxidization that results in sulfuric acid. This acid binds to water and joins up with other molecules to create aerosols that spread into a near-uniform distribution around Earth and scatter earthbound sunlight. When these aerosols are about one-half of a micron across, or a thousand times smaller than the width of a human hair, they stay aloft for roughly two years before falling to Earth. For Keith and many others, this feels like the right time span: long enough to be effective and economical, not so long that it’s irreversible.

The size distribution of the aerosols, according to Keith, remains one of the key technical uncertainties of the whole undertaking. If the aerosols get too big too quickly, then their weight drags them down to Earth early. Keith had hoped to bring light to this uncertainty in 2019, with Harvard colleague Frank Keutsch, by launching the Stratospheric Controlled Perturbation Experiment, or SCoPEx. To start with, the project aimed to float a balloon 20 kilometers over northern Sweden; release a common mineral dust; and then, using a gondola stuffed with scientific instruments and mounted with a propeller, direct the balloon on a sinuous return trip to collect samples from the plume. A test run without any material release was planned for 2021, but it was shut down by the Swedish government after the Saami Council, an Indigenous group, raised concerns.

This experience spotlighted a widespread visceral repulsion to the technology—acidifying the sky above everyone, everywhere. Keith considers this repulsion healthy. “Deliberately adding one pollutant to temporarily counter another is a brutally ugly technical fix,” he wrote in 2013. He openly disdains the cultural impatience, radiating from Silicon Valley, that mistakes technological gadgetry for well-researched and strongly reasoned solutions to serious problems. But he balances such circumspection against a single unyielding fact: The scale of the problem of climate change is proportional to the accumulated amount of carbon dioxide emitted. If the world collectively performed miracle upon miracle and brought global carbon emissions to zero in the next decade, we would not have solved the problem of climate change. Rather, we would have stopped making it incrementally worse. We would remain, for the next several millennia, citizens of a fundamentally altered planet, one transformed by the estimated 1.5 trillion tons of carbon dioxide we have emitted, nearly all of which remains in the air and oceans.

The value of SRM “is an aspect of climate that is so profoundly important because so many people don’t understand it,” says Harvard’s Jim Anderson. “In fact, a lot of people directly involved in climate and energy research don’t seem to understand it.”

However, Keith insists that SRM cannot and should not substitute for cutting emissions. This conviction underlies his greatest concern with SRM, which is unrelated to the technology itself. He fears that research into the subject and its eventual deployment could give policymakers cover to avoid the difficult and often expensive work of emissions reductions. Instead he wholeheartedly endorses the vast majority of today’s resources being spent on decarbonizing the economy. Still, he thinks SRM could be used to supplement these efforts.

Suppose, for example, the world has found and agreed to a pathway for getting net emissions to zero. It is on track to hit an average temperature of 2.5 degrees Celsius above preindustrial averages. Suppose, also, that some novel governing body has decided to use SRM to turn the global thermostat down by one degree, keeping the temperature increase to 1.5 degrees using sulfur injection in the stratosphere. This process, according to several models, would lead to roughly 10,000 additional deaths per year from ground-level pollution and ozone loss. These same models predict that the reduction in temperatures worldwide would mean one million fewer heat-related deaths every year. For every theoretical death caused by sulfate pollution, SRM would prevent one hundred.

Model after model demonstrates similarly favorable outcomes. If executed properly, “I think the scientific evidence that the benefits are quantitatively much larger than the harms, measured in human lives or environmental impacts, is strong,” Keith says. “And I think the evidence that the benefits would tend to go more to poor people in hot countries is strong.”

Keith emphasizes that, in the end, there are no risk-free choices. Adapting to our unknown future climate without the buffer of geoengineering is risky. A world in which planes release sulfur in the stratosphere is risky. The job of a voting public and its policymakers is to decide which path is more tolerable. One step removed, Keith offers his research as a way to avoid making such consequential decisions in darkness.

There is widespread recognition that roughly 20 percent of the global economy cannot be decarbonized. It’s either technologically or economically impossible on any reasonable timescale. Aviation falls into this category, as do agriculture and steelmaking. For these emissions, capturing carbon is the only way to offset what gets released.

After six years as a research scientist at Harvard, in 1999 Keith returned to Carnegie Mellon as an assistant professor and embarked on an academic exercise to estimate the cost of pulling carbon out of the atmosphere with existing technologies. The work held a kernel of promise, so he continued to refine the design, innovating within the constraints of what he could buy off the shelf. In 2009, five years after he had moved to the University of Calgary, Keith invited a fellow engineer, Robert Cherry, to tell him all the reasons his approach was foolish, to highlight all of the challenges he was overlooking. Cherry did as asked, but he also suggested there was no reason Keith couldn’t turn the idea, which was far from unviable, into a business.

Despite having no experience in the private sector, Keith followed Cherry’s advice. He founded Carbon Engineering in 2009. The company grew and, through the continual refinement of its carbon-stripping process, the technology became reasonably economical. In 2023 Occidental Petroleum acquired the company for $1.1 billion.

This was the first company in the carbon sequestration space to sell for over a billion dollars, and, as Ken Caldeira, a climate scientist at Gates Ventures who has known and collaborated with Keith for decades, points out, it made Keith a very rich man. “One of the things about David that I find super admirable is that he made a shitload of money selling Carbon Engineering and he doesn’t have to do work for anybody anymore. He could sit on the beach in Saint-Tropez if that’s what he wanted to do. He has the resources to do anything, but what he wants to do is run a research center and be a university professor.”

Not only that, but Keith and his wife have promised to donate most of their earnings to a range of concerns. They have already pledged $10 million to climate solutions across Canada.

“David is a deeply moral person, a deeply ethical person,” says Julio Friedmann, chief scientist at Carbon Direct, who first met Keith at a conference in 2003. Friedmann went on to say that Keith did not begin research into SRM because he was a mad scientist. He did it because he recognized and wanted to explore the technology’s potential to minimize human suffering. He did not start Carbon Engineering to cash in on corporate greenwashing—an accusation Keith has faced—but because he considers it a moral imperative to clean up the mess that humanity has made and preserve the world’s natural splendor. “All of his work is grounded in this moral sensibility,” Friedmann says. “To not understand that is to misunderstand the person.”

Keith now splits his time between Hyde Park, when he’s teaching, and Canmore, Alberta, a hamlet in the Canadian Rockies shadowed by sheer stone walls. He continues to rock climb. He adventures in the wilderness.

The central paradox of geoengineering is that it suggests we must alter the world to preserve it. It is easy to imagine the ways in which this goes wrong, and many people spend time doing this imagining very publicly. SRM, in particular, is readily painted as a radical and dangerous idea. But Keith would have us consider, on the other side, where we are right now, digging up hydrocarbons at an astonishing rate, burning them, dumping the product into the global atmospheric commons, changing the world in irreversible ways for a hundred generations, and leaving those generations unprotected from the very real risks of human-amplified harms: fire, drought, flood, famine. In his view, an unwillingness to face the terrifying truth of our failure and our limited options, a self-imposed blindness to technological pathways that could lead to a better future—that, too, is radical, and that, too, is dangerous.

Dylan Walsh, AB’05, is a freelance journalist based in Chicago who covers criminal justice and the environment.