If you imagine the story of the universe as a film endlessly in post-production, cosmologists would be its obsessive editors, constantly tweaking the narrative. The version they are working with is an astonishing cinematic achievement: it starts with a bang, space-time erupting out of nothing, before unfurling majestically with the formation of stars and then galaxies, sculpted by the gravitational pull of both visible matter and mysterious dark matter, all the while serenely expanding thanks to a shadowy force known as dark energy.
But it can’t be the final cut. The more we peer into space, the more it seems incomplete: the story contains niggling inconsistencies and key protagonists remain maddeningly elusive. For decades, cosmologists have been struggling to refine the script.
Now, they finally have fresh inspiration from the cosmos. A powerful telescope has mapped millions of distant galaxies to trace the expansion of the universe with unprecedented precision. What it appears to be revealing is that dark energy behaves so weirdly, it can’t be what we thought it was.
If confirmed, it is an exhilarating twist. Theorists are contemplating a complete rewrite of dark energy. How it all pans out is far from clear. But many are warming to the idea that we are about to produce a richer, more detailed cosmic story – one that looks very different from the current version.
“We’re at an interesting moment,” says Adam Riess, an astrophysicist at Johns Hopkins University in Maryland, who won a share of the 2011 Nobel prize in physics for his part in the discovery of dark energy. If someone were filming a documentary charting the making of our cosmological movie, he adds, “I would say: ‘Don’t go to the bathroom now.’”
The standard model of cosmology
Our current best picture of the origins and evolution of the universe was pieced together over the course of a century. It began in 1915 with Albert Einstein’s theory of general relativity, which describes gravity as the result of massive objects warping space-time.
At the time, the universe was thought to be static, so Einstein added a calming term to his equations called the “cosmological constant”. But in 1929, astronomer Edwin Hubble observed distant galaxies speeding away from one another, indicating that the universe is expanding and prompting Einstein to ditch his constant.
Then came the big bang theory. While it is gospel these days, it wasn’t until the 1960s that the rival steady-state theory gave way, as astronomers discovered a sea of primordial radiation left over from the big bang – the cosmic microwave background (CMB) – with properties that matched predictions.
As our ability to peer deep into space improved, the big bang theory was no longer enough. In the 1980s, astronomers found that the gravity of visible matter was insufficient to hold galaxies together or explain the formation of galaxy clusters. The fix was to invoke invisible dark matter. A decade later, observations of distant exploding stars led by Riess and his colleagues revealed, contrary to all expectations, that the expansion of the universe is speeding up. The cosmological constant was reinstated, albeit rebadged as dark energy.
And this, essentially, is the current standard model of cosmology, known as lambda-CDM. The Greek letter lambda denotes the cosmological constant and CDM stands for cold dark matter, assumed to be made of heavy, slow-moving particles. Added to general relativity and with a few key assumptions – most importantly that the universe, on average, looks the same in all directions – it offers a compelling framework for how large-scale structure formed from quantum fluctuations in the early universe through a brief burst of exponential inflation in the first moments.
The Dark Energy Spectroscopy Instrument’s new map of the universe
Lambda-CDM ranks among science’s greatest triumphs. It combines elegance with breathtaking reach, using just six parameters to describe the entire history of the cosmos, making a host of precise predictions that have been verified by increasingly exacting observations. “It has been extraordinarily successful,” says Mike Turner, a theoretical cosmologist at the University of Chicago in Illinois. “Compare it with what we had when I became a cosmologist around 1980 and, oh my God, it’s more than we could ever have imagined. It’s absolutely stunning.”
And yet, as Turner says, it is “now much less than we’re willing to settle for”. That is partly just the restless nature of science: even the most successful theories are only ever approximations of a deeper understanding and, as we stress-test them with new observations, we uncover loose ends and cracks.
In the case of lambda-CDM, the loose ends are obvious. Dark matter and dark energy were only ever placeholders: they were invoked in response to observations, but without physical explanations. Despite decades of effort, physicists have yet to directly detect dark matter particles. And while dark energy is thought of as vacuum energy, the result of quantum fluctuations in empty space, it has always been troubling from a theoretical perspective. Quantum theory predicts that its strength must be some 10120 times greater than what is required to drive the expansion of the universe we see.
“Right now, dark energy and dark matter… they’re tack-ons,” says Turner. They both serve functions. There is strong empirical evidence that they exist. “But they are just phenomenological descriptions, so they’re pointing to something more fundamental.”
The Hubble tension
Cracks have begun to appear, too, the most notorious of which has a long history, but became recognised as the Hubble tension in 2015. It is so named because two different ways of measuring the rate at which the universe is expanding, known as the Hubble constant, disagree. When cosmologists extrapolate forwards from the CMB using the current model, they get a value of about 67 kilometres per second per megaparsec. But when astronomers measure the local universe directly, using supernovae and variable stars, the value is around 73. “It’s an end-to-end test of the universe,” says Riess, who argues that the fact that the two ends don’t meet is a strong hint there is something seriously wrong with lambda-CDM.
Still, most cosmologists have been unwilling to give up on it. All the proposals made so far for how to resolve the Hubble tension undermine the existing model’s near-perfect fit to the CMB and the large-scale structure we see today. It is also possible that the measurements underlying the tension contain subtle systematic errors. The way we measure late-universe expansion in particular relies on an intricate chain of inference, each link dependent on painstaking calibration and assumptions about stars and galaxies. The suspicion is that, with more data, the tension will disappear. “There’s just too much going on there for you to say something truly definitive,” says Pedro Ferreira, a cosmologist and astrophysicist at the University of Oxford.
Riess doesn’t buy that. His measurements of late-universe expansion have been checked again and again, he points out, and nobody has found an error – even if some astronomers argue that independent distance measurements from the James Webb Space Telescope could resolve the tension. “It’s been a decade since we discovered the Hubble tension and it hasn’t gone away,” he says. “It’s only grown more pronounced.”
The real reason the community has been reluctant to move beyond lambda-CDM, Riess argues, is that scientists are loath to let go of any theory, especially such a successful one, until they have a better one. “People are uncomfortable just wandering in the wilderness.”
What we need, by that logic, are observations that more clearly point the way to something better. The good news on that front is that a new generation of telescopes designed to probe dark energy has begun to deliver in dramatic fashion, such as the Dark Energy Spectroscopic Instrument (DESI).
The DESI results
Mounted on a telescope in Arizona, DESI combines a huge mirror with 5000 robotically controlled optical fibres that automatically lock onto distant galaxies, one after the other, in quick succession – far faster than previous dark energy surveys.
Since 2021, it has been surveying millions of galaxies to gauge their redshift, or how much the light they emit has stretched due to cosmic expansion, an indicator of their distance from us. And because galaxies are at different redshifts, we can compare a characteristic spacing in their distribution – a slight preference for galaxies to be separated by a particular distance – to reconstruct how the universe’s expansion rate has changed over time.
Gabriela Secara, Perimeter Institute, NASA
To calibrate those distances, DESI has also been measuring a subtle imprint left over from the early universe, known as baryonic acoustic oscillations (BAOs). Like ripples on a pond frozen in ice, these BAOs preserve a pattern in the separation of galaxies that provides cosmologists with a “standard ruler” for measuring cosmic expansion. The idea was to produce the most accurate, precise, three-dimensional reconstruction of cosmic expansion ever made. And the latest version, released in March 2025 and based on three years’ worth of data, or 15 million galaxies, contained a bombshell that has sent shockwaves through cosmology.
When DESI researchers combined this with the latest data from supernovae – which tightly constrain the expansion of the nearby universe – and the CMB, then checked how well it all fits with lambda-CDM, they found that the current model doesn’t match up, at least not as well as one that allows the strength of dark energy to change over time. The headline finding was stark: dark energy appears to be weakening, and isn’t a cosmological constant after all.
“It was actually quite frightening,” says Will Percival, an astrophysicist at the University of Waterloo in Canada who is part of the DESI collaboration. Of course, there was a high level of scrutiny, he says. “But in many ways, this is exactly what people have been waiting for. Experiments that take us into the unknown and give us unusual, unexpected results are incredibly exciting.”
“
I like to call it beautifully bizarre “
And as if that wasn’t enough, the DESI results also suggest that in the early universe, dark energy may have dipped below the so-called phantom divide – the threshold below which its repulsive power would have been far stronger than the cosmological constant allows – before swinging back up again.
“What we’re seeing now with the DESI results, I like to call it beautifully bizarre,” says Eric Linder, a physicist and cosmologist at the University of California, Berkeley. “Not only are they off from the cosmological constant, they’re off in a way that nobody was thinking about before.” At this stage, the DESI results aren’t strong enough to claim a bona fide discovery. The analysis only favours evolving dark energy with a statistical significance of 4.2 sigma at best, some way short of the 5-sigma gold standard, making it a result that could yet vanish as more data comes in – and the phantom crossing indication is even less secure. “I’m on the fence about it,” says Ferreira, echoing the caution expressed by many in the field. “We’ve just been here so many times before.”
Even so, there are reasons to think the DESI results might be different. “It’s the first time where I’ve actually gone ‘Ha!’,” says Catherine Heymans, an astronomer at the University of Edinburgh, UK. “The method they use is one of the cleanest possible measurements of cosmic expansion we can make. It’s much harder to pick arguments with this than it is with the Hubble tension.”
Not that that has stopped people trying. In May 2025, George Efstathiou, an astrophysicist at the University of Cambridge, put out a paper claiming that the evidence for evolving dark energy is shaky for two reasons. The first is that the discrepancy with lambda-CDM only becomes apparent when the supernova data is included in the analysis. The second is that the DESI team’s statistical analysis relies on assumptions made in advance about how plausible different cosmological models are, known as “priors”, which Efstathiou argues unfairly favour evolving dark energy models.
Where there is consensus, however, is that if the DESI results do strengthen with more data, they would deal a serious blow to lambda-CDM. “In that case, it is exciting, because it means we have to think again,” says Ferreira.
In a paper published in August 2025, Riess and observational cosmologist Alexie Leauthaud at the University of California, Santa Cruz, argued that we may be witnessing the demise of lambda-CDM and that we must now prepare to move beyond it. Excitingly, for the first time in 25 years, we have a real clue as to what something better looks like.
What will replace the standard model?
Which isn’t to say it is going to be easy to figure it all out. Although the DESI results gave us a clear steer on dark energy’s physical properties, sending theorists into a frenzy, the picture of cosmic expansion they render makes it incredibly difficult to find the right formula. The simplest solution is to say that dark energy comes not from the vacuum, but is instead a kind of field similar to those that describe light or the nuclear forces. But these models require suspiciously precise fine-tuning to have dark energy grow stronger in the past few billion years, rather than some other time. More importantly, they alone can’t reproduce the phantom crossing.
Many theorists prefer to focus on models in which dark energy interacts with gravity, rather than evolving independently. The idea is that gravity begins to work differently at some point because there is a transfer of energy between ordinary matter and dark energy. “That’s how you can understand that the energy density [of dark energy] could increase and then decay,” says Alessandra Silvestri, a theorist at Leiden University in the Netherlands who has shown that such a model fits the DESI data better than lambda-CDM. “This is really the only model that seems to work.”
There are also models where dark energy exchanges energy with dark matter, allowing the latter to slowly decay into the former as the universe expands. This idea is particularly appealing from a theoretical perspective because it connects the two biggest unknowns in cosmology.
The problem with all these interacting models is that we should have seen evidence for them in existing observations of planetary orbits, for example, and we haven’t. Moreover, while it is possible that the interactions are so vanishingly small that they would have evaded detection, they could still violate the sacrosanct law of energy-momentum conservation.
What we have, then, is an abundance of ideas – none of which does the trick. “We really have no idea,” says Ferreira.
The DESI instrument on the 4-metre telescope at Kitt Peak National Observatory, Arizona
2025 The Regents of the University of California, Lawrence Berkeley National Laboratory
For Ferreira, and for Riess too, that suggests we shouldn’t just try to patch up dark energy to better match the data. Instead, we should think about what we can learn if the DESI results really are the final nail in the coffin for lambda-CDM. “We should pause a little bit and reflect,” says Riess. If we are in the first throes of another major leap in our understanding of the universe, he argues that cosmologists need to think carefully about how to navigate it – not only in terms of long-standing assumptions regarding what a better theory looks like, but also how they find it.
We may yet discover a theory as simple and elegant as what we already have. Or it might be that the best explanation is more complex – a hotchpotch including multiple dark energy fields, multiple kinds of dark matter, interactions between the two and/or a new take on gravity at cosmological scales. “This focus on elegance and simplicity, it comes from particle physics,” says Riess. “But who’s to say that it works so well at the scale of the cosmos as a whole? The universe looks pretty complicated from where I’m standing, so I think we need to be open-minded.”
Observations, as ever, will be our guide. DESI is still collecting data, with another data release expected in 2027. But cosmologists are also expecting big things from the European Space Agency’s Euclid space telescope and the Vera Rubin Observatory in Chile, both of which started releasing data last year. That should give us greater confidence in the emerging picture of expansion, or not. But it will also allow us to see it from previously unexplored redshifts.
Ferreira is less optimistic. In a 2025 paper, he and his colleagues argued that because cosmological surveys can only probe a limited period of the universe’s expansion history, many different theoretical models can produce nearly identical behaviour over that interval. As a result, Ferreira reckons that even with all the new data coming in, “we will be left with a large family of models which are essentially observationally indistinguishable from the point of view of the cosmological data”.
The danger is that we will end up exactly where we were with the Hubble tension – an impasse in which many cosmologists aren’t prepared to let go of lambda-CDM because they don’t trust the data that would break it without a better theory in place, and little prospect of finding that theory any time soon. Riess describes this scenario as “Kuhnian purgatory”, in reference to the philosopher Thomas Kuhn’s ideas about how scientific progress plays out, and worries it will lead to inertia. “Trying to pull the sword out of the stone, it’s hard work… and you might not get a lot of papers out of it. But let’s not forget that sword is still stuck in that stone.”
Time for a paradigm shift
That said, he suggests the problem lies not with the data to come. Rather, it is that the community places too much weight on a model developed before new data came along, and not enough on the data itself. Whenever a tension arises within lambda-CDM, he says, the inability to explain it is held up as evidence against the new observations – which explains why the community fixates on unknown errors and more measurements only breed more doubt. “When you live with a standard model for 20 years, a lot of people have spent most of their career with it,” says Riess. “Even the idea that this might not be the whole story, it’s jarring.”
Maybe this is just the nature of paradigm shifts. They will always be marked by conflict, and lambda-CDM will not go gently into the night. But that’s not necessarily a problem. “You want the defenders to look for anything that looks a little suspicious in the data. You also want the revolutionaries, the people who are willing go beyond what we already have,” says Linder. “The back-and-forth, while it may look antagonistic, it’s actually healthy.”
Indeed, the fact that cosmologists are gearing up for a fight might itself indicate we really are poised for another revolution. The one thing we can say for certain is that, after a long period of harmony, cosmology is entering an era of tensions that make it a whole lot more interesting. “We’re looking forward to all this new data, which, I think, will thrill us all,” says Linder. “It’s just an incredibly exciting time.”
