Rubin’s cosmos work continues as U.S. grants wobble

U.S. universities – A new generation of American researchers is pressing ahead on tools that are already built and on ideas that can’t wait—while uncertainty over federal funding slows the flow of money to labs and trainees. From cosmology shaped by the Vera C. Rubin Observatory

When Sydney Erickson started graduate school at Stanford in 2021, she knew only that she was going to be a physicist. She rotated through particle physics and cosmology—until she began hearing the buzz about a giant camera being built on campus for the Vera C. Rubin Observatory.

“I was drawn toward that community,” Erickson says, remembering the people involved with the telescope. “Rubin drove me into cosmology, I would say.” The telescope turned on last summer in the Chilean Andes.

Now Erickson is finishing her doctoral degree and developing new ways to measure the universe’s accelerating expansion. Her work focuses on huge astrophysical objects such as massive galaxies, whose gravity magnifies the light from more distant targets. By studying the arrival time of light from gravitational lenses—so-called because gravity bends and focuses light—cosmologists can calculate how much the universe expanded during different time intervals. It’s a complex measurement that demands lots of images captured repeatedly. which Rubin provides. and deep finesse. which Erickson and her computer models bring.

The Rubin observatory itself—an $800-million project designed and built largely with federal money—signals what’s still pushing U.S. science forward. Federal funding has long been a primary engine for research and development in the U.S. growing from $21.3 billion in 1956 to $156.1 billion in 2024. adjusted for inflation. according to the National Science Foundation.

But the money that keeps today’s labs running is not arriving at the pace researchers say they need.

Since last January. when President Donald Trump took office for the second time. he began slashing programs and implementing major funding cuts to dozens of government departments and agencies. Morale among scientists is low. and government scientists have left their jobs in droves; about 10. 000 science and technology Ph.D.s have exited the government since Trump’s reelection. Many graduate students, other than Erickson, have not been as fortunate. Erickson will start a postdoctoral fellowship at the University of California, Los Angeles, this summer.

So far 2026 is. unexpectedly. a good time to be an astrophysicist in the U.S. at least for people who can hitch their work to instruments already built and paid for. Some telescopes recently turned on. including Rubin. and others are set to launch this year. including the Nancy Grace Roman Space Telescope.

But the picture darkens if Congress moves toward the Trump administration’s proposed 55 percent cut to the NSF budget. In that scenario, grants and early-career programs for people who want to use those telescopes would drop dramatically or face elimination.

“There is this huge momentum behind what has been the U.S. science enterprise.” —Mari Ostendorf, University of Washington

Even as researchers described that momentum, they also described a familiar tension: success depends on what’s already in place, while new funding uncertainty makes planning harder.

To understand how different fields are coping. MISRYOUM reached out to researchers and leadership at institutions that receive major funding from the NSF and the NIH. Deans and vice provosts were asked about the mood in their scientific communities. how funding has changed. what they’re hopeful about. and what worries them most.

Even in the most optimistic corners, administrators kept returning to one theme—uncertainty. Michael Graham. interim associate vice chancellor for research and an engineering professor at the University of Wisconsin–Madison. said that even when budgets are technically implemented. labs may not feel the benefits.

“Although there are bright spots, I think it’s important to understand that even though a reasonable budget was implemented [this year], we are not seeing it on the ground,” he says. Grants from both the NSF and the NIH have been slow to reach the labs that rely on them.

Mari Ostendorf—vice provost for research at the University of Washington—had been hearing similar messages since attending a meeting of the Association of American Universities this past February. Senior research officers at dozens of universities described how funding awards were down substantially compared with amounts in previous years. Ostendorf said the University of Washington was also seeing funding losses. but that “good projects are continuing to produce good science.”.

“A lot of these things happen because of the initial investments of the federal government,” Ostendorf says, pointing to the runway that earlier funding created.

At Johns Hopkins, Peter Armitage, a condensed matter physicist, framed it more starkly. The foundation built over decades of growth and investment is strong—computers are more capable, instruments more advanced, and collaboration across disciplines and institutions is easier than ever.

“One has to separate public perception of science. the actual doing of science and the outcomes from the doing of science. ” Armitage says. “For right now, for the last one, we are just drinking from the fire hose. In biology, in physical sciences, there’s just gobs of things going on. We almost can’t keep up with how exciting it is.”.

Still, the “fire hose” can’t run forever if money and hiring stall.

In many disciplines, researchers said new computing tools are helping them do more with what they have. Graham’s work in rheology—the physics of how matter responds to forces—illustrates why. Rheology can reveal how materials, including mRNA vaccines packaged inside lipid nanoparticles, move through the body. Graham said the challenge is that computers aren’t great at simulating fluid flows. largely because scientists don’t have enough basic physics knowledge for AI systems to model that flow accurately.

Graham and chemical engineer Matthew Helgeson of the University of California, Santa Barbara, developed a technique that uses x-ray scattering measurements of flowing fluids containing polymer molecules. They feed those measurements into a machine-learning algorithm to generate more accurate models.

“This is a really great big-data example in my field of soft matter,” Graham says. “We can learn the governing equations for how the microstructure evolves in time as the fluid is flowing, and we can move away from simple models.” Practical uses could include designing more efficient solar cells.

In gerontology, machine learning is changing what questions researchers ask. At the University of Texas at Austin. early-career scientist Elizabeth Muñoz is working with older people to build a fuller picture of what happens in an aging brain. Instead of conducting research primarily in clinical settings, Muñoz surveys people at home and in their communities. She gives participants frequent. short cognitive tests. records their stress levels. and notes whether they live alone or with loved ones.

Muñoz correlates cognitive test results with demographic information, environmental and neighborhood information, and other data. She has already published research showing that cohesive, supportive neighborhoods are linked to better cognitive function. Researchers then feed the data into computers. where the results help paint a more complete picture of a person’s dementia risk.

Karen Fingerman, who directs the Texas Institute on Dementia, Aging and Longevity, called Muñoz’s approach revolutionary.

“We used to come up with the research problems based on the literature we were reading. The new perspective is to go out there and ask people what’s happening,” Fingerman says. “Let’s not just bring you in and put you in a machine. let’s not have some neurologist give you a test. but let’s go to where you are. and let’s use your smartphone and design something new that will assess something in your life.”.

Another U.T. Austin researcher, Stephanie Grasso, is studying whether bilingualism and language fluency can protect people from dementia. In one study on language use. Fingerman and colleagues found that people with deeper thinking patterns tend to use third-person and plural first-person pronouns (“they. ” “we. ” “us”). while people with less complex thinking use just first person. The finding emerged when transcripts were fed into a computer and a machine-learning algorithm analyzed study participants’ speech patterns.

“Saying ‘I went to the store’ is much simpler than explaining what someone else did,” Fingerman says, adding that the result was unexpected. “We would not have found that five years ago, because we would not have looked for third-person pronouns, but we found it with machine learning.”

For Alzheimer’s and other dementia research, the funding picture is also getting reshaped from outside the federal pipeline. Although federal grants are down. state funding for research on Alzheimer’s and other forms of dementia ballooned starting in 2024. according to the Alzheimer’s Impact Movement. a nonprofit advocacy group.

At the University of California, San Diego, researchers are working on a campus-wide effort to produce an Alzheimer’s vaccine, says Corinne Peek-Asa, an epidemiologist and the university’s vice chancellor for research and innovation.

“Let’s try a lot of things and fail along the way, and boy, are we going to learn a lot in the process,” Peek-Asa says.

Infectious-disease research is also expanding at universities, even as federal programs are disrupted. In early 2025, officials led by Secretary of Health Robert F. Kennedy, Jr. eliminated more than 800 grants focused on HIV research, vaccine hesitancy, and transgender health. The administration pushed thousands of federal scientists to quit or take early retirement, including at the NIH.

By mid-May 2026, the Centers for Disease Control and Prevention, the nation’s public health agency, had been without a confirmed leader for all but one month of Trump’s second term.

That atmosphere hasn’t stopped universities from trying to solve problems quickly. At the University of Washington, researchers developed CandyCollect, a new way to diagnose strep throat. If not treated quickly, strep infection can lead to serious complications. Instead of a swab of the back of a patient’s throat. the technique uses a lollipop of sorts: a flavored candy head collects a child’s saliva in a grooved spiral. The candy is engineered to dissolve over a set time so that when the flavor is gone. the sample is ready. Lab technicians can then test the lollipop for Streptococcus bacteria.

One bright spot that many researchers pointed to is protein science—work that ranges from understanding how amino acid chains fold into shapes to creating engineered molecules for medicine and even environmental cleanup.

Protein research sits at a fruitful nexus between health and AI, both described as priorities for the Trump administration. In 2024, University of Washington scientist David Baker won the Nobel Prize in Chemistry for computer systems that can design proteins. For his breakthrough, Baker used AI diffusion models to create atomic arrangements quickly. The university’s Institute for Protein Design recently released one of those models as open-source software. freely available to any scientist. Baker said that when anyone can use the same code, advances can benefit researchers everywhere.

His team has been looking at applications such as building proteins able to break down plastics.

A couple thousand miles southeast, at U.T. Austin, computational protein engineer Danny Diaz leads the Deep Proteins Group within the AI Institute for the Foundations of Machine Learning (IFML), funded by the NSF. Diaz’s team has patented half a dozen novel proteins whose design was hastened by AI.

“In protein engineering, nature is really good at giving us an MVP,” Diaz says. “An AI model is useful because [it] can speed up evolution, so we can create proteins [whose] function is on par with human needs.”

During his doctoral research at U.T. Austin in 2020. Diaz worked down the hall from a separate team that designed a shelf-stable version of one of the most consequential proteins in recent years: the coronavirus spike protein. Diaz said that by making a few modifications. researchers could turn the protein into biotech that is useful—such as making it stable in the lab. Those modifications enabled scientists to produce a vaccine in record time.

Diaz said that work drew attention from Adam Klivans, who directs the IFML. Diaz recalled that Klivans said, “Wow, proteins seem important,” but that he didn’t speak the same language. Now the two work together to design proteins using generative software techniques originally developed for language models or image production.

The vaccine space has also been hit by funding fights. In 2025 and 2026, the Trump administration moved to cut funding for vaccine research at the NIH and the CDC. Last June. Kennedy fired all 17 members of a key vaccine advisory panel at the CDC. and his handpicked replacements have since been blocked by federal courts. Kennedy also ordered sweeping changes to the number of shots recommended routinely for American children. a move tied up in litigation. In April 2026. Kennedy blocked $600 million in funds for a State Department–controlled program that provides vaccines for children in developing countries.

Even so, protein research is moving forward. Diaz’s group created an open neural network called MutCompute. a machine-learning platform that determines mutations to optimize many types of proteins. including enzymes. Enzymes speed up chemical reactions. Diaz said that MutCompute applied just three mutations to a natural enzyme called PETase to create an enzyme that can break down the plastics in single-use water bottles. The team is also designing a protein that can bind to a type of cancer receptor. possibly making it easier to target cancer cells.

“It’s like looking for a needle in 1,000 haystacks, so that is where AI is very, very helpful,” Diaz says.

For all the new methods, officials said a central bottleneck remains federal disbursement. Peek-Asa said the biggest challenge is “the unpredictability.” Officials across the country have said the NIH and the NSF are not disbursing grants as quickly as usual.

With federal timelines slipping, state and private money has increasingly stepped in. Research funds from state governments and private donors “mushroomed” last year. even in states where such investment may come as a surprise. In Texas. state legislators passed laws that dissolve faculty representation. restrict free speech on campuses by limiting protests. prohibit diversity. equity and inclusion initiatives. and require faculty to certify they are not indoctrinating students.

At the same time. last November voters approved a $3-billion. 10-year project to study Alzheimer’s disease. Parkinson’s disease. and other ailments of aging. In California. state lawmakers are debating a University of California–sponsored ballot issue to support a $23-billion endowment for research across the state. Peek-Asa said California has a history of stepping up when research is threatened. citing previous efforts to fund stem cell research after federal bans in the early 2000s.

Private funding for major projects is not new, but researchers said the scale keeps climbing. In 2008. the Rubin observatory received a $30-million gift from Bill Gates and Microsoft software architect Charles Simonyi for construction of its enormous mirrors. Last August. philanthropist Penny Knight and her husband. Nike co-founder Phil Knight. gave $2 billion to Oregon Health and Science University’s Knight Cancer Institute. described as the largest-ever private donation to a U.S. university. The University of Notre Dame. a private university. is adding faculty positions this year to the Stavropoulos Center for Complex Quantum Matter. funded by former Dow Chemical chair Bill Stavropoulos and his wife. Linda Stavropoulos.

Across the country in 2025, 186 institutions received a windfall of nearly $7.2 billion from philanthropist MacKenzie Scott. Since 2020. Scott has donated more than $1.1 billion to dozens of historically Black colleges and universities. tribal colleges. two-year schools. and institutions that promote college access.

Nobody is sure how long these windfalls will last. And private dollars can’t replace federal money that supports science year after year. Armitage said some universities are moving forward with hiring because of private funds, but “if that is all there is, that’s not sustainable.”

Still, federal money is flowing toward projects aligned with the Trump administration’s priorities. Even as the White House called for dramatic cuts to scientific research funding this year. congressional appropriations remained mostly stable. says Massimo Ruzzene. senior vice chancellor for research and innovation at the University of Colorado Boulder.

“All of us were shaken up,” Ruzzene says. “Now we have some relief in sight, given the funding levels, so I think there is cautious optimism.”

In Texas. the IFML’s funding was renewed for $20 million from the NSF for 2026 to study the fundamental algorithms and architecture that drive generative AI. At the University of Texas at Austin, the Texas Advanced Computing Center hosts the fastest supercomputer system in U.S. academia. Later this year. it will bring online a new supercomputer called Horizon. described as the largest open-use science supercomputer in the country.

Graham said AI, quantum science, and fusion technology are bright spots. Scientists working on projects in those areas might have an easier time securing funding. he said. because they match the administration’s priorities. But he is not convinced AI is as promising for science as some people suggest.

“To my mind, that is good and bad,” Graham says. “Those tools are going to help us largely in fields where we know a lot of the physics, where we have a lot of the data.”

In California. where lawmakers are trying to plug gaps in federal funding. Peek-Asa said the University of California system—which includes 10 campuses such as U.C.S.D. U.C. Irvine. and U.C.L.A.—is still positioned to work on subjects including AI and quantum technology. which the Trump administration has said are high priorities. “I think there is a lot of concern that all the agencies are kind of moving to the same priorities because they are safe. ” she said.

Meanwhile, U.C.S.D. is working with industry. In fall 2024 the university launched a new Fusion Engineering Institute. hiring professors and creating courses to steer people into possible fusion careers. Fusion energy became newly attractive after a major DOE breakthrough in 2022. when scientists were finally able to produce more energy from a nuclear fusion reaction than the laser energy used to drive it. While commercializing fusion is a long way off. researchers told MISRYOUM that they view fusion as a net-positive area partly because the Trump administration considers it a major priority.

Ruzzene said administrators and scientists should work on explaining why basic research matters—because the results take time, and because the value is not always obvious to people outside the lab.

“That’s on us,” Ruzzene says. “We have to do a better job of doing that than we used to.”

Before speaking with MISRYOUM. Ostendorf prepared a list of projects at the University of Washington. some of which are delivering new findings. She pointed to work ranging from the diagnostic lollipop to citizen scientists monitoring the health of baby crabs in the Salish Sea. She said the university has suffered more than other schools from funding cuts. but she welcomed the chance to talk about positive news.

Across universities, many experts linked their worry to a broader sense that higher education and research-driven work are under attack. For administrators—many of whom are still active researchers—the stress is manifold.

“What I need to do in my administrative role is make sure people don’t give up. Just keep submitting proposals,” Ruzzene said. “Money has been appropriated, and at some point [the Office of Management and Budget is] going to have to spend it, and we are well positioned.”

Armitage said he sees more entry-level jobs in physics now than there were a few years ago. Educators are getting jobs. but so are people with bachelor’s-level education. hired to work in machine learning. quantum materials. and other fields. “But I don’t want to come off as flippant saying the world is great,” Armitage added. “The job market is not great overall.” He described “a lot of storm clouds” and said the system would weather coming squalls if grants start coming back out later this year and funding returns to normal. “But if it doesn’t happen, then we are going to see a generational collapse in science.”.

Erickson, the Stanford graduate student and soon-to-be U.C.L.A. postdoc, said she knows nothing is guaranteed.

“One thing I think about is how lucky I am to be able to study fundamental science and make a living off it. I get to think about the universe expanding and ask, ‘What’s happening, what’s our fate?’” she says. “I feel really lucky to be able to keep doing the science for as long as I can.”

Vera C. Rubin Observatory NSF funding NIH funding astrophysics cosmology gravitational lenses machine learning dementia research Alzheimer’s vaccine protein design AI diffusion models open-source software CandyCollect strep throat