Dark Energy Is Weakening — And the Universe May Be Slowing Down
The last lecture Marina gave on the certainty of endings was on a Tuesday in March. She stood at the podium of Hall 104 at the University of Arizona, three hundred undergraduates in folding chairs before her, and explained — with the practiced warmth of someone who had given this talk thirty times before — that the universe would expand forever.
She said forever the way a doctor says terminal. Gently. As though the diagnosis itself were a kind of mercy.
"You can be sad about it," she told them, "or you can be free about it. Either way, the math doesn't change. Dark energy makes up sixty-eight point five percent of everything that exists. It is the dominant force in the universe, and it is pushing everything apart. Galaxies. Stars. You. Eventually — in a trillion years, give or take — all the lights in the sky will drift beyond the horizon of any telescope we could ever build. The universe will be dark. Not black. Dark. There's a difference. Black is a color. Dark is an absence so total it has no name."
A girl in the third row took notes. A boy near the back checked his phone. Marina watched both of them with equal tenderness, because she understood that the ones who took notes were trying to hold on to something, and the ones who checked their phones were trying to let something go, and both impulses were the same impulse, and the name for that impulse was being alive on a planet that would one day be gone.
She had given this lecture since 2015. She would give it again. The Lambda CDM model — the standard model of cosmology, Lambda for dark energy as a cosmological constant, Cold Dark Matter for everything else — was one of the most rigorously tested frameworks in the history of science. It had passed every observational test. The cosmic microwave background. The large-scale structure of the universe. The acceleration of expansion discovered in 1998 by two teams who shared a Nobel Prize for it. The model was not just confirmed. It was settled. You could build careers on settled. You could build meaning on it too — the meaning was bleak, but at least it was certain, and certainty had its own strange comfort.
Marina went home that night and ate leftover pasta standing at the kitchen counter. Her daughter, twelve years old, was at her ex-husband's this week. The apartment was quiet in the particular way that quiet can be loud — the silence of a space that knows it is missing someone. She poured a glass of wine she wouldn't finish, opened her laptop, and pulled up the latest DESI collaboration data dump.
She was not supposed to look at it yet. The collaboration used a blinding technique — results were hidden from the researchers until the analysis was finalized, a method borrowed from particle physics to prevent unconscious bias from steering the analysis toward what people wanted to find. The blinding was a form of intellectual honesty so deeply institutional that it had become almost ritualistic, like genuflection before a door. You closed your eyes. You did the work. Then you opened them and looked at what was actually there.
Marina knew the blinding was appropriate. She also knew she had been a cosmologist for twenty-two years, and that the urge to peek was not dishonesty but hunger, and that the two were easy to confuse.
She did not peek. She closed the laptop. She went to bed. She did not sleep well.
The thing about dark energy is that nobody knows what it is.
This is not a minor epistemological inconvenience. It is the central drama of modern cosmology, played out in telescope domes and conference rooms and the pages of journals that almost no one outside the field will ever read. Dark energy is not dark, exactly. It is not energy, exactly, if by energy you mean something with a temperature, a spectrum, a fingerprint. Dark energy is a name for a phenomenon: the observed acceleration of the expansion of the universe, which requires an energy density of roughly 0.003 joules per cubic kilometer of space. That is an almost inconceivably small number. A firefly's worth of energy spread across a volume larger than Earth. And yet it is the dominant force in the cosmos. It is, in the language of physics, everything.
The leading model — Lambda, the cosmological constant — treats this energy as a fixed property of space itself. Empty space has energy. That energy causes expansion. The expansion creates more empty space, which has more energy, which causes more expansion, endlessly, forever. The model is elegant. It fits the data. It has been the intellectual furniture of cosmology for two decades.
Andrei Cuceiu, a postdoctoral researcher at the Flatiron Institute and a member of the DESI collaboration, had given an interview after the 2025 results that Marina had read so many times the words had begun to feel like scripture. "We are in the business of letting the universe tell us how it works," he said, "and maybe the universe is telling us it's more complicated than we thought it was."
More complicated. That was the diplomatic way of saying it. The precise way was this: the data from DESI's map of fifteen million galaxies and quasars — built using five thousand fiber-optic cables that captured light from five thousand galaxies simultaneously, mounted on a four-meter telescope on a mountain in Arizona called Kitt Peak, which in the language of the Tohono O'odham Nation is called I'oligam Du'ag, the place of the mountain — the data from all of it suggested that dark energy is not constant. It appears to be weakening. The acceleration of the universe is decelerating.
The standard ruler that DESI uses is called baryon acoustic oscillation — BAO, in the shorthand of the field. In the first fraction of a second after the Big Bang, the universe was a hot plasma, and sound waves rippled through it, leaving pressure imprints in the distribution of matter. Those ripples froze into place as the universe cooled, leaving a characteristic scale of roughly 150 megaparsecs — about 500 million light-years — imprinted in how galaxies are distributed today. By measuring this frozen ripple at different distances, DESI could trace how the expansion rate of the universe had changed over eleven billion years.
And what DESI found was that the expansion rate at different epochs was not quite what Lambda CDM predicted. Not dramatically wrong. Not obviously wrong. Wrong in the way that a sentence is wrong when a single word is off — the grammar still works, the meaning is still mostly there, but something in the rhythm has shifted, and you can hear it if you listen.
The DESI team checked the signal against itself. They blind-tested. They ran the analysis a hundred different ways. The signal did not go away.
Then came Union3.
The Supernova Cosmology Project had spent years assembling the largest standardized dataset of Type Ia supernovae ever compiled. These are the standard candles of cosmology — explosions so consistent in their intrinsic brightness that they can be used as distance markers across the universe, like a chain of lighthouse beacons across an ocean of space. The 1998 discovery that won the Nobel Prize used about fifty supernovae. Union3 used 2,087. They came from 24 different surveys, calibrated across datasets using Bayesian hierarchical modeling — a statistical technique that can account for unknowns by constraining them with other unknowns, a mathematical way of saying we don't know everything, but we know enough to say something.
The Union3 results, published in July 2025, pointed in the same direction as DESI. The expansion history of the universe over the last seven billion years did not quite fit Lambda CDM. The preferred model — not proven, not confirmed, but preferred by the data — was one in which dark energy evolves over time, growing weaker.
When two independent methods using two independent datasets reach the same conclusion, you pay attention. The statistical significance varied depending on how the data were combined — 2.8 sigma here, 3.8 sigma there, 4.2 sigma in the analysis that used the most assumptions. The gold standard for a discovery in physics is 5 sigma, a one-in-3.5-million chance of a statistical fluke. None of the individual combinations had reached that threshold. But the direction was the same across all of them, and the direction was wrong for Lambda CDM, and that mattered more than any single p-value, because science is not really about p-values. It is about the universe telling you something you did not expect to hear.
Saul Perlmutter, who shared the 2011 Nobel for discovering the acceleration of expansion — who had spent his career building the edifice of dark energy cosmology — was quoted as suppressing premature elation. The signal could still go away with better data. Many three-sigma results in physics had faded away. But he was not dismissing it. None of them were.
Marina read the papers. She read them twice. She read the preprints. She attended the seminars. In October 2025, she flew to Phoenix for the DESI collaboration meeting, and she sat in a conference room with four hundred other cosmologists, and she watched the presentation of the combined DESI + Union3 results, and she felt something move through the room that was not quite excitement and not quite fear but something in between — the particular feeling of people who have dedicated their lives to understanding the structure of reality, and who have just been told that the structure might be different than they thought.
After the lecture, a student approached her.
His name was Eli. He was twenty-three, finishing his master's thesis on primordial black holes, which were also, tangentially, a possible explanation for dark energy if dark energy turned out not to be a cosmological constant. He had dark eyes and a nervous way of holding his hands, and when he spoke, he spoke quickly, as though afraid the words would escape if he didn't get them out fast enough.
"Dr. Vance," he said. "The lecture. About the universe expanding forever."
"Yes."
"Is that still — I mean, is that still the thing?"
Marina looked at him. She understood what he was really asking. He was not asking about the Lambda CDM model. He was asking whether the future he had been told was inevitable was still inevitable. He was asking whether the sky he would grow old under was still the sky he thought it was.
"The data is interesting," she said carefully. "There are tensions with the standard model. But we don't have a discovery. We have a hint."
"But it could change," Eli said. "The fate of the universe. It could actually be different."
"It could."
Eli nodded slowly. He did not look relieved. He did not look frightened. He looked, if anything, awake — the way someone looks when they have just been told that the map they have been navigating by might be wrong, and that the territory they are standing in might be something other than what they thought, and that the act of standing here, now, might be more significant than anyone had told them.
"Thank you," he said, and walked away.
Marina watched him go. She thought about her daughter, who was twelve and who sometimes asked her questions about the universe that she answered honestly and then wished she had answered differently — not because the honest answer was wrong, but because the honest answer was always too big, and a twelve-year-old did not need the full weight of cosmic uncertainty dropped on her at bedtime. She thought about the girl in the third row who had taken notes. She thought about the boy in the back who had checked his phone.
She thought about the fact that in 1998, when the acceleration of the universe was first discovered, the scientists involved did not believe their own results. They checked the data. They checked it again. They looked for instrument errors, calibration errors, statistical errors. They were sure something was wrong. Something was wrong. The universe was accelerating, and they had found it, and the Nobel Prize they eventually received was not for confirming what they believed but for believing what they found.
The Vera C. Rubin Observatory, formerly known as the Large Synoptic Survey Telescope, was coming online in Chile. The Nancy Grace Roman Space Telescope would follow from orbit. Within a few years, the datasets would be ten times larger. The signal — if it was real — would either solidify into a five-sigma discovery or dissolve into noise. Either outcome would be valuable. Either outcome would tell the universe something true about itself.
Marina knew this. She had lived through enough scientific cycles to understand that the most exciting moment in any discovery is not the moment of confirmation but the moment of not-knowing — the liminal space when the signal is strong enough to matter but not strong enough to trust, when the universe is whispering and you are leaning in with everything you have, and you do not yet know if what you are about to hear is a secret or just the wind.
She went back to her office. She made coffee she wouldn't drink. She opened her laptop and looked at the DESI data file — the one she had not been supposed to look at — and she saw the plot that showed the expansion history of the universe, the data points and the Lambda CDM prediction line, the slight but persistent gap between them, the whisper that would not go away.
She thought about what it would mean if the gap was real. Not the physics — she could calculate the physics, and the physics was interesting but not the point. She thought about what it would mean existentially. If dark energy was weakening, the expansion could slow. It could stop. It could reverse. Everything that had drifted apart since the Big Bang — galaxies, stars, planets, the atoms in your bones — could begin moving back toward each other, drawn by a gravity that would grow stronger as the universe contracted, until all of it — every love, every war, every city, every thought anyone had ever had — was compressed into a single point of infinite density, the same kind of nothing-that-is-everything that the universe had exploded out of fourteen billion years ago.
The Big Crunch. The ending that was also a beginning. The loop.
Marina sat with this for a long time.
Then she closed the laptop, went to the window, and looked up at the sky. It was still light. The stars were not visible yet. But she knew where they were — fifteen million of them in the DESI map alone, and billions upon billions more beyond that, all of them drifting away from each other at this moment, every second, carried by an energy that might be eternal or might be fading, and might end in cold silence or might end in fire, and nobody knew.
Nobody knew.
She thought: that is the most honest thing the universe has ever said to us.
She thought: we have been wrong before, and we will be wrong again, and the act of being wrong is how we learn where the edges of the map are.
She thought about Eli, and the girl with the notebook, and her daughter asking about what happens to the stars, and the Tohono O'odham people who called this mountain by a different name and understood the sky not as a prediction but as a story — something that moved, that changed, that could not be pinned down.
She thought about Saul Perlmutter, suppressing premature elation. About Andrei Cuceiu, saying the universe might be more complicated than we thought. About the five thousand fibers on Kitt Peak, each one capturing light from a galaxy that had taken billions of years to reach the mountain, and how all of that ancient light was being used to ask a question about the future that no one alive would see answered.
She thought about the word forever, and how we use it like it means something simple, and how it has never meant anything simple, and how the not-knowing is not a failure of science but its highest purpose — the proof that the universe is still in motion, and that we are part of that motion, and that the story is not over.
The sky darkened. The first stars came out.
Marina watched them.
We are in the business of letting the universe tell us how it works. And maybe the universe is telling us it's more complicated than we thought it was.
The stars burned. The expansion continued. Whether it would slow, or stop, or reverse — whether the universe was dying by cold or by fire or by some third thing no one had imagined yet — was not known.
The uncertainty was, in its own terrible way, the most beautiful thing about it.
Everything was still in motion. And so were we.
---