In 2024 a shockwave rippled through the astronomical world, shaking it to the core. The disturbance didn’t come from some astral disaster at the solar system’s doorstep, however. Rather it arrived via the careful analysis of many far-distant galaxies, which revealed new details of the universe’s evolution across eons of cosmic history. Against most experts’ expectations, the result suggested that dark energy—the mysterious force driving the universe’s accelerating expansion—was not an unwavering constant but rather a more fickle beast that was weakening over time.
The shocking claim’s source was the Dark Energy Spectroscopic Instrument (DESI), run by an international collaboration at Kitt Peak National Observatory in Arizona. And it was so surprising because cosmologists’ best explanations for the universe’s observed large-scale structure have long assumed that dark energy is a simple, steady thing. But as Joshua Frieman, a physicist at the University of Chicago, says: “We tend to stick with the simplest theory that works—until it doesn’t.” Heady with delight and confusion, theorists began scrambling to explain DESI’s findings and resurfaced old, more complex ideas shelved decades ago.
In March 2025 even more evidence accrued in favor of dark energy’s dynamic nature in DESI’s latest data release—this time from a much larger, multimillion-galaxy sample. Dark energy’s implied fading, it seemed, was refusing to fade away.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Soon afterward, however, Daniel Green, a physicist at the University of California, San Diego, took to social media to argue over the DESI team’s preferred interpretation of the data.
“I’m particularly skeptical of DESI’s press release,” Green says. “The tendency should be to say, ‘Hey, why don’t we explore all the possible interpretations?’ DESI didn’t do that many analyses.” The situation, Green says, is akin to looking for a lost set of car keys in a dark parking lot—but only where the light is bright: “When all you look under is one lamppost, you only see what you find there.”
Other explanations exist for DESI’s measurements, Green says, and not all of them require the cosmos-quaking prospect of an evolving dark energy. His preferred model instead invokes the putative decay of another mysterious aspect of cosmology, dark matter—thought to be a substance that gravitationally binds galaxies together but otherwise scarcely interacts with the rest of the universe at all. Yet his and other alternative proposals, too, have drawbacks, and the resulting scientific debate has only just begun.
Constant Cosmology
The standard cosmological model at the heart of all this is known as “LCDM.” The “CDM” component stands for “cold dark matter,” and the “L” stands for the Greek letter “lambda,” which denotes a constant dark energy. CDM is the type of dark matter that best accounts for observations of how galaxies form and grow, and—until DESI’s proclamation suggested otherwise, that is—a constant dark energy has been the best fit for explaining the distributions of galaxies and other patterns glimpsed in large-scale cosmic structures. “Once they had this constant, everything snapped into place,” Green says. “All of the issues that had been around for 20 years that we’d been hoping were just small mistakes were really resolved by this one thing.”
But dark energy’s constancy has always been more of a clever inference rather than an ironclad certainty. DESI is an effort to clarify exactly what dark energy really is by closely monitoring how it has influenced the universe’s growth. Since 2021 the project has been meticulously measuring the motions and distributions of galaxies across some 11 billion years of cosmic time.
DESI’s data on galactic motions come from measurements of redshift, the stretching out of galaxies’ emitted light to the red end of the spectrum by the universe’s expansion. And its tracing of spatial distributions emerges from spying enormous bubblelike arrangements of galaxies thought to have formed from more primordial templates, called baryon acoustic oscillations (BAOs). BAOs are essentially ripples from giant sound waves that coursed through the hot plasma that filled the early universe, which astronomers can glimpse in the earliest light they can see, the big bang’s all-sky afterglow known as the cosmic microwave background (CMB). The waves’ matter-dense crests sowed the seeds of future galaxies and galaxy clusters, while galaxy-sparse voids emerged from the matter-poor troughs. Combined with CMB data as well as distance-pegging observations of supernovae, DESI’s measurements offer a reckoning of the universe’s historic growth rate—and thus the action of dark energy.
A virtual “flight” through a portion of the largest 3D map of the universe to date, using data from the Dark Energy Spectroscopic Instrument (DESI).
DESI co-spokesperson Nathalie Palanque-Delabrouille, a physicist at Lawrence Berkeley National Laboratory, recalls the private December 2023 meeting where she and the rest of the DESI team first learned of the project’s early results. Up until then, the researchers had worked on blinded data, meaning the true values were slightly but systematically altered so as to ensure that no one could deliberately or inadvertently bias the ongoing analysis to reach some artificially preordained result. These blinded data showed a huge divergence from LCDM. But when the real data were unveiled, “we saw all the points came very close to LCDM, and that was initially a huge relief,” she recalls. That alignment suggested “we did things right.”
Those feelings quickly changed when the group noticed a small, persistent deviation in DESI’s estimate for the value of lambda. Still, there was a considerable chance that the results were a statistical fluke. But in DESI’s latest results, which were posted to the preprint server arXiv.org last March and incorporated much larger and richer data sets, the statistical robustness of the unexpected lambda value soared, and most talk of flukes dwindled.
Theorists could scarcely contain their excitement—or their profound puzzlement. The results rekindled preexisting ideas about dynamic dark energy first formulated decades ago, not long after dark energy’s discovery itself in 1998. One popular theory posits a fifth fundamental force in addition to the known four (electromagnetism, gravity, and the strong and weak nuclear forces), emerging from some as-yet-undiscovered dark matter particle that can influence dark energy. Frieman says the data from DESI is so precise that if this particle is the correct explanation, physicists already know its crucial parameters.
Constrained by the DESI data, Frieman says, the best-fitting model that would support this “fifth force” hypothesis “tells us that this [hypothetical] particle has a mass of about 10–33 electron volts.” To put that into perspective, this means such a particle would be 38 orders of magnitude lighter than an electron—which, Frieman notes, is “by far the lightest stable particle we know of that doesn’t have zero mass.”
But while some theorists used DESI’s data to revive and sharpen intriguing theories of yesteryear, Green and others issued a warning. The problem: an evolving dark energy would seem to defy well-founded physical principles in other cosmic domains.
Null and Void
The first major point of controversy involves something called the null energy condition, under which—among other things—energy can’t propagate faster than light. If circumstances were otherwise, then perilous paradoxes could emerge: time machines could violate causality, matter could repel rather than attract, and even spacetime itself could be destabilized. Theorists have mathematically proven the condition’s apparent necessity in numerous circumscribed scenarios within quantum and relativistic domains—but not for the universe at large. Appealing to this sort of theoretical incompleteness, however, “is like a lawyer saying there’s a loophole,” Green says. “Most physicists would say that’s totally crazy.”
A discovery that something in the universe violates the null energy condition would be groundbreaking, to say the least: a more impolitic term would be “nonsensical.” This astounding violation is exactly what Green and others say most of DESI’s analyses are showing, however. On this point, several theorists push back. The controversy goes all the way down to the foundations of modern cosmology, centering on a parameter unceremoniously known as w(z).
In 1917 Albert Einstein first introduced lambda as a way to ensure that a static universe would pop out of his equations. But after work led by Edwin Hubble proved the universe was expanding, Einstein abandoned his fudge factor (even calling it his “greatest blunder”). It wasn’t until the late 1990s, when astronomers found that the universe’s expansion wasn’t constant but in fact accelerating, that lambda once again returned to theoretical prominence. This time theorists interpreted it to represent the magnitude of the universe’s dark energy density, a constant that doesn’t change with time.
But if there’s one thing modern cosmology has shown, it’s that little, if anything, about the universe is ever so neat and tidy. So, despite a lack of evidence, theorists of the time reimagined LCDM as w(z)CDM, where w(z) is a time-varying term representing the ratio of dark energy’s pressure to its energy density. When w(z) has a value of exactly –1, w(z)CDM is equivalent to LCDM. For w(z) greater than –1, the universe’s dark energy dilutes over time, consistent with DESI’s findings. On the other hand, w(z) less than –1 leads to devastating consequences: dark energy’s pressure overpowers its density, ultimately causing everything from galaxies all the way down to atoms to be ripped apart—a “big rip” that violates the null energy condition and would seemingly doom the universe to a violent death.
The DESI group collaboration’s March preprint includes a graph that shows w(z) with values below –1 for later epochs in the universe’s history, seemingly validating the criticisms of Green and others. But all is not as it seems. Such criticisms “draw the wrong conclusions,” says Paul Steinhardt, a cosmologist at Princeton University.
That’s because in a second graph in the DESI paper, w(z) never crosses the critical –1 line. The difference: despite DESI’s curved data, the first chart uses a simple line fit for w(z). Steinhardt and Frieman both say that because of the poor fit, the linear w(z) isn’t physically meaningful. Researchers merely find it convenient for comparing different dark energy models and experiments.
The second graph shows a curved fit for w(z) that more closely matches the data. It rolls down to, but never crosses, the critical –1 value, consistent with a weakening dark energy that would avoid the universe ending in a big rip.
But Gabriel Lynch, a Ph.D. student at the University of California, Davis, who has an alternative explanation for the DESI data, says that even if any of DESI’s w(z) estimates are physical, coaxing out a theory to support them leads to incredibly fraught circumstances. “This is saying something weird,” Lynch says. “It’s not impossible, but maybe it would be good to look into some alternatives.”
Negative Neutrinos?
Whether or not DESI’s results would violate the null energy condition, everyone agrees on another problem. Models that accommodate a changing dark energy inevitably conclude that a class of tiny fundamental particles known as neutrinos have a negative mass. Yet multiple generations of empirical experimentation have indisputably shown that neutrinos do have mass. Frieman suggests that something else, perhaps an unknown particle, might be mimicking a negative-mass neutrino.
But a new approach by Lynch and his thesis advisor Lloyd Knox, detailed in a preprint that was posted to arXiv.org in March, sidesteps this “negative neutrino” problem altogether. If some of the mass in the universe somehow disappeared over time, its influence on DESI’s data would be the same as a weakening dark energy—without necessitating a negative mass for neutrinos. Although physicists have good reasons to believe that certain seemingly stable subatomic particles could contribute to this notional effect by decaying over time, this process is thought to be far too slow to account for DESI’s observations. For instance, experiments have shown the proton to be so stable that its half-life must be at least a hundred trillion trillion times the age of the universe. But no one knows what the half-life of putative particles of dark matter would be. So, Lynch asks, what if dark matter has a half-life of roughly a billion years? Fast forward about 14 billion years to today, and some would have decayed into dark radiation, erasing the heavy matter signal.
If the idea holds true, DESI’s data might be a way to find the exact value for neutrino masses as well as for dark matter particles, which would be a big deal. “That is a breakdown of LCDM that we totally expected,” Green says. “And we were just waiting to detect it.”
The Truth Is Out There
Owing to dynamic dark energy’s paradoxes, “you really need to explore every alternative explanation [for the results], because evolving dark energy is the absolute last one that I would be willing to believe,” Green says.
Despite such strong words, all parties caution that this debate is still in its early days. “This is only the first round of the fight,” Steinhardt says, and no model currently explains all of DESI’s results. More data are needed, especially from even bigger and better cosmic surveys by planned next-generation telescopes. And, naturally, more analyses are needed, too, before the community can reach any consensus. Whether a resolution comes from dynamic dark energy, dark matter decay or something entirely different, the LCDM model has seemingly been stretched to its breaking point. Every reasonable explanation for DESI’s data involves new, scarcely explored physics. “They are all exotic models. We’re beyond LCDM both ways,” Palanque-Delabrouille says. “We just want to know the truth.”