For decades the Hubble Space Telescope defined the frontier of deep-space imaging. Over ten days at Christmas 1995 it stared at a single patch of sky, taking 342 separate exposures totalling more than 100 hours to produce the Hubble Deep Field North. ¹ But Hubble reached a hard ceiling: it worked in visible light, and the light from the most distant objects is stretched so far by the expansion of the universe that it drops out of the visible range entirely, requiring infrared eyes to see it. ¹

The James Webb Space Telescope was built for exactly that infrared frontier. It launched on 25 December 2021 and, over its first years, has been pointed at the earliest galaxies, at planets around other stars, and at the growth of black holes. ¹ This report weighs its discoveries against the graded evidence — separating the firm results from the contested ones, and testing the claim, repeated often, that JWST has upended our understanding of the cosmos.

One technical point frames everything that follows. A bigger mirror does not automatically mean a sharper picture. Because resolution depends on the ratio of wavelength to mirror size, an object observed at a 2-micron wavelength yields an image no sharper from JWST than from Hubble, despite Webb's far larger mirror. ¹ JWST's advantage is its reach into the infrared and its speed, not simply its size.

The clearest triumph of JWST's early-universe program is distance. Within its first six months, the JADES survey spectroscopically confirmed four galaxies at redshifts between 10.3 and 13.2. ¹ Redshift is a stand-in for distance and age: the higher the number, the further back in time we are seeing. GN-z11, long a candidate, was pinned down at a definitive redshift of 10.6034. ¹ After nearly two years, the tally had grown to more than 14 spectroscopically confirmed galaxies beyond redshift 10, with roughly 400 more candidates awaiting confirmation — a consistent growth curve from a single survey team. ¹

The record-holder is JADES-GS-z14-0, confirmed as the most distant known galaxy — though two authoritative sources give slightly different redshifts for it: NASA 14.32 (+0.08/-0.20), the Pontifical Academy of Sciences 14.43, with NASA's error bars not including the higher figure and the discrepancy unresolved. ² The NASA figure corresponds to less than 300 million years after the Big Bang. ¹ NIRSpec, Webb's spectrograph, stared at it for almost ten hours in January 2024 to nail the measurement, breaking the previous record of redshift 13.2. ¹ An independent arXiv analysis reports two luminous galaxies confirmed near redshift 14, representing additional spectroscopic confirmations at about 300 million years after the same epoch, demonstrating that JWST is finding multiple galaxies at this extreme distance rather than a single outlier. ¹

All of this rests on JWST's speed. Its First Deep Field, an image of the galaxy cluster SMACS 0723, combined exposures totalling 12.5 hours and reached infrared depths beyond Hubble's deepest fields, which took weeks. ¹ That comparison is not a clean like-for-like — it sets a 12.5-hour total exposure against Hubble's calendar time of weeks — but the direction is unambiguous: JWST reaches fainter, more distant objects far faster than its predecessor. ¹

Where the deep-field work is about distance, JWST's exoplanet work is about chemistry — reading the composition of alien atmospheres by watching starlight filter through them. Here the telescope logged genuine firsts. It made the first detection of carbon dioxide in an exoplanet atmosphere, at the hot giant WASP-39 b, and the first constraint on photochemistry — chemical reactions driven by starlight — outside the Solar System, through the molecule sulfur dioxide. ² The carbon dioxide result is corroborated by more than one Tier-1 source. ² JWST also observed a Jupiter-mass planet with an atmosphere dominated by carbon rather than the expected water, methane and carbon dioxide, an unusual composition flagged by University of Chicago researchers. ¹

The most publicised exoplanet claim, though, illustrates why caution matters. In April 2025 a Cambridge-led team reported dimethyl sulfide — a molecule made on Earth largely by marine life — in the atmosphere of K2-18 b, at 3-sigma confidence, meaning only about a 0.3 percent chance the signal was a fluke. ¹ Three months later a NASA-led reanalysis of the same data downgraded the signal to roughly 2.7-sigma, below the 5-sigma bar the scientific community demands for a firm detection. ² The two teams differ not on the data but on how to model it — different atmospheric assumptions and treatment of systematic error. ² And even a solid detection would not prove life: NASA JPL's Renyu Hu notes that photochemistry alone, with no biology involved, can produce detectable dimethyl sulfide and related organosulfur molecules. ¹

A further complication sits under all such measurements. Espinoza and colleagues confirm that stellar surface blotchiness — the transit light source effect — distorts the transmission spectra JWST records, a problem worst for planets orbiting small M-dwarf stars, and one predicted before launch. ¹ That is precisely the class of system many biosignature searches target.

Open: Whether higher signal-to-noise JWST observations of K2-18 b can push any dimethyl sulfide signal to the 5-sigma threshold remains untested in the evidence here.

Beyond individual galaxies and planets, JWST has reshaped the picture of how the first massive structures grew. Its most striking cosmological finding concerns black holes. Quanta Magazine reports that JWST found active supermassive black holes in early galaxies 10 to 100 times more abundant than expected from counts of mature quasars. ¹ A CEERS analysis led by Larson in March 2023 identified a black hole betraying itself through a broad hydrogen emission line at redshift 8.7 — about 0.57 billion years after the Big Bang — the most distant active black hole then known. ¹ Wikipedia reports that data published in 2024 place the most distant known black hole inside GN-z11, at around 1.6 million solar masses. ¹

That last figure rests on a single Tier-3 source and should be treated as reported rather than confirmed. ¹ The abundance and record-distance findings, by contrast, come from a reputable outlet quoting named experts. ¹ The consistent thread is that black holes grew large earlier than formation models anticipated.

JWST's imaging power has also produced landmark public images at the same time. Its Stephan's Quintet mosaic, released on 12 July 2022, is the largest image the telescope has produced, covering about one-fifth of the Moon's apparent diameter, built from over 150 million pixels and almost 1,000 separate image files. ¹ Its October 2022 NIRCam view of the Pillars of Creation captured jets from newly forming stars as red spots at the pillars' edges. ¹

The most consequential question JWST raised was whether it had broken cosmology itself. The alarm came from early galaxy masses. A Nature Astronomy study by Boylan-Kolchin, reported by McDonald Observatory, found six early massive galaxy candidates seen 500 to 700 million years after the Big Bang, each exceeding 10 billion solar masses. ¹ To build so much stellar mass so fast, Boylan-Kolchin said, these galaxies would have had to convert nearly 100 percent of their available gas into stars — against a typical ceiling of about 10 percent. ¹ Astrobites summarised the mood: such galaxies were 'way larger than what ΛCDM predicts,' throwing 'a curveball' at the standard model. ¹

That framing travelled further than the science supported. Eric Lerner's August 2022 article claiming JWST images disprove the Big Bang was rated mostly false by a fact-check drawing on named experts. ¹ And the professional response pointed toward measurement rather than cosmology. An arXiv analysis found that adding JWST/MIRI data reduced derived stellar masses by 0.4 dex for most high-redshift galaxies — a downward revision of roughly a factor of two and a half in the very quantity that had driven the crisis. ¹ Separately, the same modeling work showed that JWST galaxy populations between redshifts 7 and 14 are naturally reproduced within standard ΛCDM when galaxy formation is modeled with the UniverseMachine code applied to the Uchuu simulation. ¹

A third line of evidence cuts the other way on the exotic explanations. A Sabti, Muñoz and Kamionkowski paper accepted at Physical Review Letters in 2024 showed that Hubble's ultraviolet luminosity-function data cannot accommodate the power enhancements that would be needed to explain a genuine mass excess — meaning any cosmological modification large enough to fit the JWST claims would clash with well-established Hubble data. ¹ Together these results suggest the tension arose from how masses were estimated, not from a failure of the model. ² Whether that reconciliation is final, or whether some residual tension survives independent replication, is the open part of the story.

Open: Whether independent simulation groups replicate the UniverseMachine/Uchuu reconciliation, and whether reanalysis of the Boylan-Kolchin sample with MIRI data brings its specific galaxies into ΛCDM ranges, is not settled in the evidence here.

No instrument this complex operates without wear, and JWST's technical record is a mix of resilience and real loss. The deployment went almost flawlessly: by 8 January 2022, 14 days after launch, all deployments were complete, with only minor anomalies quickly resolved and no impact on science. ¹ An accurate Ariane 5 injection left so much fuel that NASA estimates the mission can run for more than 20 years — a single-source figure, but from the operating agency itself. ¹

The most serious hardware event was a micrometeoroid strike. Between 22 and 24 May 2022, an impact caused significantly higher wavefront error on primary mirror segment C3 — a segment that had already been hit once before. ³ That event is corroborated across several sources. ³ Over 11 months, NASA's wavefront sensing identified 14 micrometeoroid impacts on the primary mirror in total. ¹ These are documented degradations, not merely cosmetic ones, though the mission continued operating through them. ³

There was also an outright capability loss. SpaceNews reports that a MIRI medium-resolution spectroscopy mechanism began showing increased friction, first noticed on 24 August 2022, and that NASA stopped using that observing mode on 20 September 2022. ¹ That single-sourced but specific account records a real gap in JWST's instrument suite. ¹ Set against the record galaxies and first atmospheric detections, these are the costs of operating a delicate observatory in a hostile environment.

Open: Whether the MIRI medium-resolution spectroscopy mode was later restored, and how much science it cost, is not addressed by the sources here.

The evidence supports several competing explanations for JWST's most disputed findings, and they deserve to be tested against one another rather than resolved by assertion.

On the early galaxies, one reading holds that JWST broke ΛCDM cosmology. It rests on the Boylan-Kolchin candidates each exceeding 10 billion solar masses, the near-impossible gas-to-star conversion they would require, and the 'curveball' language of early coverage. ² A competing reading holds that the tension was an artifact of mass estimation. It rests on the 0.4-dex downward revision when MIRI data are added, the successful reproduction of JWST populations within ΛCDM using UniverseMachine and Uchuu, and the Physical Review Letters result that Hubble data rule out the cosmological modifications the crisis would demand. ² The evidence currently tilts toward the second reading: the same observations that looked anomalous are reproduced within the standard model once systematics are handled. ¹ What would settle it is independent replication of that modeling and a MIRI-corrected reanalysis of the original Boylan-Kolchin sample.

On K2-18 b, one reading treats the dimethyl sulfide signal as evidence of biology, resting on the April 2025 3-sigma detection. ¹ This is the weakest of the hypotheses on the current evidence. The July 2025 reanalysis pulled the signal down to about 2.7-sigma, below the detection bar, and non-biological photochemistry can produce the same molecules. ² Only higher signal-to-noise observations reaching 5-sigma, combined with modeling that rules out photochemical production, could discriminate.

On early black holes, the evidence points more firmly toward a genuine puzzle. Their abundance runs 10 to 100 times above predictions from mature quasar counts, with a confirmed active black hole at redshift 8.7. ¹ No claim in the record contradicts the observation itself; what remains open is whether standard formation physics — heavier seeds, faster accretion — can accommodate it. Here the discovery is real; the explanation is unsettled.

The evidence forces a split verdict on JWST's discoveries. Two categories are firm. First, the distance records: JADES-GS-z14-0 is the most distant known galaxy, confirmed spectroscopically over nearly ten hours of observation, part of a population of more than 14 confirmed galaxies beyond redshift 10. ² Second, the exoplanet firsts: carbon dioxide detected in an exoplanet atmosphere, and photochemistry constrained beyond the Solar System for the first time. ² These stand as landmark results on confirmed, in some cases independently corroborated, evidence.

The most publicised potential breakthrough is weaker than its coverage suggested. The K2-18 b dimethyl sulfide signal sits at about 2.7-sigma after reanalysis, below the threshold for a firm detection, and has a plausible non-biological explanation. ² Presenting it as a discovery of life would overstate what the data support.

The cosmology scare has largely, though not entirely, resolved. The claim that JWST disproved the Big Bang was rated mostly false, and the more serious early-massive-galaxy tension is reproduced within ΛCDM once mass estimates are corrected. ³ The evidence supports treating this as a measurement problem that the scientific process addressed, while noting the reconciliation still awaits broad independent replication.

One genuine anomaly survives: early supermassive black holes appear 10 to 100 times more abundant than models predicted. ¹ Speculatively — and labeled as such — this may prove the most scientifically consequential JWST result, precisely because, unlike the mass tension, it is a robust observation still lacking an agreed explanation; the reasoning is that no claim in the record contradicts the count, whereas the galaxy-mass anomaly was undercut by revised measurements. Finally, the instrument itself is durable: 14 micrometeoroid strikes and a lost MIRI mode notwithstanding, it carries fuel for more than two decades of work. ²