The Galaxies That Shouldn't Exist: Why JWST Is Quietly Breaking Cosmology
There's a story that a small team of astronomers at the Space Telescope Science Institute has become quietly famous for.
In late 2022, the James Webb Space Telescope pointed at a patch of sky no bigger than a grain of sand held at arm's length. The team downloaded the resulting images the next morning. In one corner, they found something small, red, and very faint. They ran the redshift math — the way astronomers estimate how far away a galaxy is by how much its light has been stretched by the expansion of space. The answer came back at z ≈ 14.
Which is to say: 13.5 billion years ago. Roughly 300 million years after the Big Bang.
They double-checked. Then triple-checked. And then — the story goes — one of the senior astronomers walked out of the room, sat on a bench in the hallway, and said out loud to no one: "This is not supposed to be here."
That galaxy is now called JADES-GS-z14-0. It's real. It's big. It's older than our theory of the universe allows. And in the four years since, JWST has found dozens more like it.
What JWST actually does, in plain English
Every telescope on Earth is a time machine. When light from a distant galaxy hits your telescope, that light started its journey millions or billions of years ago. The farther away the galaxy, the older the light. Look far enough, and you're looking at the young universe — before galaxies were fully formed, before most of the elements existed, when the entire cosmos was still figuring itself out.
The problem is that visible light from very early galaxies gets stretched by cosmic expansion, sliding into the infrared before it reaches us. Regular telescopes can't see it. You need an infrared telescope, cold and precise, sitting far from Earth's heat. That's JWST.
Since 2022, JWST has been doing what nothing before it could: looking at the universe as it was ~200 to 500 million years after the Big Bang. That's the "cosmic dawn" — a period we had theories about but almost no direct data on.
We expected to find a few small, dim, gassy protogalaxies. Baby galaxies. Wispy things still gathering material. That's what the standard cosmological model — called ΛCDM — predicted.
We got something else.
What "shouldn't exist" actually means
The first surprise was mass. Multiple JWST galaxies at z > 10 (redshifts corresponding to under 500 million years after the Big Bang) appear to have as much stellar mass as galaxies do today. That's a problem, because in the standard model, matter needs time to clump together, form stars, and build up mass. There simply wasn't enough time by then for that many stars to have formed.
The second surprise was the "little red dots." Starting in 2023 and getting more numerous ever since, JWST has cataloged hundreds of very compact, very red objects at high redshift. Their spectra suggest they're either extremely dense clusters of very old stars — which shouldn't exist that early — or unusually massive supermassive black holes with accretion disks that shouldn't be that big yet either.
The third surprise was consistency. This isn't one weird galaxy. It's dozens. Roughly 10-30% of the earliest galaxies JWST samples look "impossible" by the numbers we thought were right.
The theory of the universe we've spent 40 years building predicts a specific rate of galaxy assembly. The observations don't match. Not by a factor of two. In many cases, by a factor of ten.
What's actually going on, according to the three main camps
There are three competing explanations. All three are real, active areas of research in 2026.
Camp 1 — Our galaxy models are wrong, not the universe. In the early universe, stars may have formed much more efficiently than they do today — with far less material lost to feedback and heat. The math of how clumps of gas turn into stars might just look different in ultra-dense early conditions. If this is right, the observations are legitimate but the theory needs updating on stellar formation efficiency, not on the fundamental structure of the cosmos.
Camp 2 — The redshift measurements are being fooled. Some of these "z ~ 14" galaxies may not actually be that far back. Certain kinds of dust and gas can mimic high-redshift spectra. There's a small but real chance we're systematically overestimating distances for a subset of the objects. This is the boring answer, but it happens in astronomy more often than you'd think.
Camp 3 — The fundamental cosmology is wrong. This is the exciting one. Modifications like early dark energy, primordial black holes seeding galaxy formation earlier than expected, or alternatives to ΛCDM (like MOND or fuzzy dark matter) could explain why galaxies got big fast. If Camp 3 is right, the past 40 years of cosmology needs revision. Not thrown out — revised.
Nobody knows yet which camp will win. All three are being tested with more JWST time and follow-up observations from Chile-based telescopes and radio arrays like ALMA.
Why this matters beyond "cool space picture"
The reason JWST's findings are a bigger deal than headlines suggest: cosmology is the closest thing physics has to a "successful settled science." The standard model of cosmology has predicted, with startling accuracy, the temperature of the cosmic microwave background, the ratio of hydrogen to helium in the universe, the arrangement of galaxies across billions of years of history. It works.
Except now, at the earliest epoch we can observe, it doesn't.
This is the pattern that broke Newtonian physics in the early 1900s. Newton worked — for centuries — except for a few weird observations at extreme scales (Mercury's orbit, the speed of light, black-body radiation). Those weird observations turned into general relativity and quantum mechanics. The whole edifice of modern physics is downstream of taking anomalies seriously.
JWST may or may not be the next Mercury orbit. But the pattern is worth respecting.
What we'll know by 2030
Three concrete things.
The Roman Space Telescope launches in 2027. It's a wide-field survey instrument — designed to catalog vast numbers of high-redshift galaxies, so we know whether the "impossible" galaxies are a rare weird sample or a genuine population.
The Extremely Large Telescope (ELT) sees first light around 2028. With its 39-meter primary mirror, it can measure redshifts of some of JWST's targets far more precisely, ruling in or out the "we got the distance wrong" theory.
Follow-up spectroscopy from JWST itself is nearly complete on the current sample. Deep spectra will nail down whether these objects are really old, really big, and really far — or whether some of them are dust-reddened intermediate-redshift interlopers.
By 2030, we'll either have confirmed that early-universe physics needs a real revision, or we'll have quietly patched the models with better stellar-formation efficiency and moved on. Either way is interesting. One of them is historic.
The reflection
The thing I love most about JWST is that it wasn't sent up to find anomalies. It was sent up to test our cosmological model — to fill in the picture of the young universe with data instead of speculation. The scientists who built it fully expected it to confirm what we already believed.
Instead, in its very first year of operation, it started to gently but firmly break the model.
That's how physics almost always advances. Not through a lone genius sitting alone in an attic. Through a telescope that a thousand engineers assembled correctly, a set of astronomers who checked the data honestly, and a decade of quiet nights running spectra until the pattern became undeniable.
Somewhere between now and 2030, someone will publish a paper that either resolves this cleanly or forces cosmology to add a genuinely new ingredient to the recipe. When they do, remember it started with a very small, very red dot in a patch of sky no bigger than a grain of sand.
The universe, it turns out, is quieter than we thought — and stranger.