One of the most important scientific institutions in the United States is tucked away in the foothills outside of Boulder, Colorado. The National Center for Atmospheric Research, or NCAR for those who work in atmospheric science, has been building the tools, models, and computers that make it possible to predict the weather for more than 60 years. You can keep track of hurricanes on your phone, cities in the west are told to keep kids inside because of wildfire smoke, and commercial flights are rerouted because of turbulence. A lot of that is due to work that was done here. Now, Trump’s government wants to break it up.
The news didn’t come from a formal policy document or a hearing in Congress. Instead, it came from a post on X by the director of the Office of Management and Budget, who called NCAR a source of “climate alarmism.” Another letter from the National Science Foundation in January asked for outside bids to take over parts of the center’s infrastructure. Then, in February, NSF confirmed what many researchers had been most afraid of: NCAR’s main supercomputer, Derecho, which can do almost 20 quadrillion calculations per second, would be given to a third-party operator who has not been named. The letter didn’t give a date. There was no word on who would take over. There was no guarantee that the about 4,000 scientists who are currently using Derecho would still be able to use it.
It’s still not clear what the administration wants this to look like when everything is over, and that’s what’s making people nervous.
Derecho is housed in a building in Cheyenne, Wyoming, and it’s more than just a fast computer. Real infrastructure is this: a national resource that all universities and research teams can use. They could never afford to build or maintain it on their own. Scientists can get time on the National Science Foundation by applying for it. They then use that time to model wind energy systems, simulate solar storms, and improve the forecast algorithms that NOAA relies on. Notably, NOAA recently decided to use a framework created by NCAR researchers to improve its next-generation weather models. This is true even though the federal budget process often treats them as two separate line items.
The part that worries researchers the most is separating the supercomputer from the scientists who built and run it. Julie Lundquist, an atmospheric science professor at Johns Hopkins, made it clear: being able to walk down the hall and talk to a world expert on turbulence in the atmosphere is not a luxury. This is the right way to do science. If you give Derecho to a third-party operator, that relationship goes away completely.
There’s a bigger issue at play here than just this research project. Accurate weather modeling isn’t just a school project; it affects decisions in emergency management, agriculture, construction, aviation, and the sea. Farmers plan when to plant based on weather forecasts. Turbulence models that NCAR researchers helped make are used by airlines to change the routes of flights. Allied pilots in the Pacific during World War II were caught off guard by strong high-altitude winds that no one had yet mapped. People now know a lot about those air currents, called jet streams, in large part because organizations like NCAR studied them for decades.
Some people in charge in Colorado think that the move has less to do with scientific priorities and more to do with political pressure. In Denver, there is a theory that the administration is going after NCAR at least in part to put pressure on Democratic Governor Jared Polis over a request for clemency from a convicted election denier. It’s possible that’s not true. It’s the kind of detail that’s hard to forget once you hear it, though.
There’s no doubt that scientists are not going to be quiet. In a letter to NSF, former NCAR director James Hurrell and dozens of his coworkers said that any plan to break up the center is “fundamentally not in the nation’s interest.” The community of people who study the atmosphere has come together in an unusually quick and united way. It is the job of these researchers to argue about cloud parameterization schemes all day long. You should pay attention when they agree on something so strongly.
We still don’t know if that opposition will change anything in Washington. There aren’t many signs that the administration will change its mind. In a letter to his staff, NCAR director Everette Joseph admitted the hard truth: he doesn’t know who will take over the supercomputer, when it will happen, or what access researchers will have afterward. He said it was bad news. That seems like too little of an answer.
