The phrase “impossible early galaxies” has survived JWST, but its meaning has changed. As of September 2026, the strongest conclusion is not that ΛCDM cosmology has been falsified. It is that galaxy formation during the first billion years was substantially faster, burstier, and more efficient than our pre-JWST models expected—and a few parts of the puzzle have actually become more uncomfortable.
The original problem predates JWST. Steinhardt and colleagues called attention in 2016 to galaxies that seemed to have assembled too much stellar mass too early relative to the available dark-matter halo population. JWST then made the problem much harder to dismiss, because it found unexpectedly many luminous galaxies at z>10, and spectroscopy established that at least some really are that distant rather than merely dusty lower-redshift impostors. For example, JADES spectroscopy confirmed galaxies at z=10.3–13.2, corresponding to roughly 300–450 million years after the Big Bang. These very earliest confirmed systems were generally young and relatively modest in stellar mass, around 10^7–10^8\,M_\odot, so they themselves were not “Milky Ways 300 Myr after the Big Bang.”
That was the first major correction to the sensational 2022–23 story: some of the initially inferred stellar masses were too high. Photometric redshifts, strong emission lines, dust, AGN light, and assumptions about stellar populations could all masquerade as enormous old stellar populations. So the very strongest early claims—“galaxies as massive as the Milky Way only a few hundred million years after the Big Bang”—mostly softened.
But the underlying anomaly did not disappear.
What JWST has established rather securely is that there are too many luminous galaxies at z\sim10–15 compared with many pre-JWST expectations. A 2025 review described the bright 10<z<15 galaxy population as overabundant relative to Hubble-era extrapolations. JWST is therefore telling us that early galaxies can turn gas into luminous stars remarkably rapidly.
Even more strikingly, by z\sim6–7—when the Universe was still under a billion years old—we now find galaxies that are not merely forming stars furiously, but have apparently already formed enormous stellar populations and begun shutting down. JWST spectroscopy has found post-starburst/quiescent systems with inferred stellar masses \log M_*/M_\odot\gtrsim10.6, whose main stellar episodes occurred around z\sim7–9. That gives nature only a few hundred million years to assemble tens of billions of solar masses in stars and then quench them.
And there is a second “impossible” population: early black holes. JWST has uncovered numerous compact broad-line AGN whose black holes appear unusually massive relative to their host galaxies. There is growing evidence that at least some may be undergoing short periods of super-Eddington accretion, which could solve part of the timing problem. Recent work also argues that some black-hole masses may have been overestimated. Thus the black-hole crisis is real as a population puzzle, but its magnitude is still moving.
There is, intriguingly, a brand-new twist. A Nature Astronomy result published only weeks ago argues that some massive galaxies may contain many more low-mass stars than normally assumed. Since those faint stars contribute lots of mass but little light, conventional stellar-population modelling could underestimate rather than overestimate stellar masses—by as much as a factor of several in some systems. If this result generalizes to genuinely high-redshift galaxies, it would make the “too much mass too early” problem worse again, not better. It is not yet established that the same IMF applies at cosmic dawn, so this is a warning rather than a settled conclusion.
So I would separate what we know from what remains interpretation:
- Certain: substantial galaxies existed within ~300–400 Myr of the Big Bang; spectroscopy has confirmed them.
- Very likely: the abundance of bright galaxies in the first ~500 Myr is higher than many pre-JWST models predicted.
- Certain by ~1 Gyr: some galaxies had already accumulated 10^{10}–10^{11}M_\odot-scale stellar populations and, astonishingly, some had already quenched.
- Very likely: star formation was extraordinarily efficient and bursty, with rapid gas inflow and short intense starbursts playing a major role.
- Still uncertain: precise stellar masses, the stellar IMF, dust content, AGN contamination, halo masses, duty cycles and selection effects.
- Not established: that ΛCDM itself is wrong.
The distinction is important. ΛCDM gives us the dark-matter scaffolding; prescriptions for gas cooling, star formation, feedback, dust, black-hole growth and stellar populations are additional astrophysics. Most current tensions can potentially be accommodated by changing the latter without changing cosmology. Indeed, the 2025–26 reviews increasingly frame the issue as “galaxy formation was much more violent and efficient than anticipated,” rather than “the Big Bang model has failed.”
But there is one thing I would no longer say: “JWST’s impossible galaxies were just an early calibration mistake.” That explanation has become untenable. Individual objects came down in mass; the population-level surprise remained.
Conceptually, the interesting shift is this:
Before JWST: dark-matter halos form → gas settles → galaxies gradually assemble.
After JWST: halo appears → gas catastrophically collapses → enormous starburst → enrichment, black-hole growth and sometimes quenching—all compressed into a few hundred million years.
So “impossible” increasingly means impossibly fast, rather than literally impossible.
And there is a beautiful deeper question here. The universe seems to have reached a surprisingly high level of organization very shortly after it became capable of making organization at all: metals, dust, structured galaxies, massive black holes, and even dead galaxies all appear remarkably quickly. JWST has moved the frontier from “When did the first galaxies appear?” to something much stranger:
Why was cosmic evolution so efficient almost immediately after it began?
That, I think, is the surviving Impossible Early Galaxy Problem.
