Webb Space Telescope Discovers Massive Galaxy That Does Not Rotate

When galaxies form, they spin. Gas and dark matter collapse under gravity, conservation of angular momentum kicks in, and the resulting structure rotates — sometimes at hundreds of kilometers per second. That is how it works in every major simulation, every textbook, and nearly every observation to date. So when a team led by UC Davis research scientist Ben Forrest pointed the James Webb Space Telescope at a massive galaxy from the early universe and found it barely rotating at all, the result warranted attention.

The galaxy, designated XMM-VID1-2075, sits at a redshift of z = 3.449, meaning its light has traveled roughly 12 billion years to reach us [1]. We are seeing it as it appeared when the universe was approximately 1.8 billion years old. It contains about 330 billion solar masses' worth of stars — several times more than our own Milky Way — and it is no longer forming new ones, producing less than one solar mass of new stars per year [2]. In virtually every respect, it resembles the giant elliptical galaxies that populate the centers of galaxy clusters in the modern universe. Except the modern universe has had 13.8 billion years to build such objects. XMM-VID1-2075 had less than two.

The findings were published May 4, 2026 in Nature Astronomy [3].

What the Observations Show

Forrest's team used JWST's Near-Infrared Spectrograph (NIRSpec) in integral field unit (IFU) mode to map how material moves inside XMM-VID1-2075 and two comparison galaxies of similar mass and age [2]. The IFU mode breaks a galaxy's image into a grid of tiny spatial elements called spaxels — each 0.1 arcseconds across, corresponding to about 730 parsecs at this redshift — and captures a full spectrum at every point. This allows astronomers to measure the velocity of stars at different locations within the galaxy, building a two-dimensional kinematic map.

For XMM-VID1-2075, those velocity maps showed stellar velocities offset by less than 100 km/s everywhere across the galaxy, while the random velocity dispersion — the speed at which stars move in disordered directions — measured approximately 387 km/s within the effective radius and peaked near 500 km/s at the center [2]. The resulting spin parameter, λRe = 0.123, places the galaxy firmly in the "slow rotator" category [2].

Source: Forrest et al. 2026, Nature Astronomy

Data as of May 7, 2026CSV

The contrast with the two companion galaxies is stark. XMM-VID3-2457, at a similar redshift of z = 3.489 and stellar mass of 180 billion solar masses, showed a spin parameter of 0.671 — a clear, organized rotation. XMM-VID3-1120, with 300 billion solar masses, registered an intermediate λRe of 0.296 [2]. "This one in particular did not show any evidence of rotation, which was surprising and very interesting," Forrest said [1].

Why a Non-Spinning Galaxy Is Unusual at This Epoch

Slow rotators are not unknown. In the local universe, roughly 15 percent of massive early-type galaxies are classified as slow rotators — systems where random stellar motion dominates over organized rotation [4]. They are dispersion-supported rather than rotationally supported, meaning their structure is held up by the chaotic motion of billions of stars moving in all directions rather than by the collective centrifugal force of an orderly spin.

But slow rotators in the local universe are understood to be the end products of a long evolutionary process. They tend to sit at the centers of galaxy clusters, with stellar masses exceeding 2 × 10¹¹ solar masses, and their kinematic state is thought to result from repeated dry mergers — collisions between gas-poor galaxies that gradually randomize stellar orbits over cosmic time [4][5]. The MAGPI survey confirmed that the slow rotator population was largely in place by z ≈ 0.3, or about 3–4 billion years ago [5].

Source: Schulze et al. 2018 (Magneticum); Forrest et al. 2026

Data as of May 7, 2026CSV

In the Magneticum Pathfinder cosmological simulation, the fraction of slow rotators among quiescent galaxies increases from about 8 percent at z = 2 to 30 percent at z = 0 [6]. At z ≈ 3.4 — the epoch of XMM-VID1-2075 — the same simulation predicts that only about 3 out of 35 quiescent galaxies (roughly 8.6 percent) would qualify as slow rotators [2][6]. XMM-VID1-2075 falls within the predicted range, but it is among the very first observationally confirmed examples at such high redshift.

The Merger Hypothesis: A Cosmic Head-On Collision

If slow rotators typically require billions of years of mergers, how did XMM-VID1-2075 get there so fast? The research team favors a specific scenario: a single major merger between two galaxies whose angular momenta were oriented in nearly opposite directions [1][7].

The evidence for this comes from the galaxy's morphology. JWST imaging reveals asymmetric, low surface brightness features — an excess of light extending to one side of the main body [2]. "For this particular galaxy, we see a large excess of light off to the side," Forrest explained. "And so that's suggestive of some other object which has come in and is interacting with the system and potentially changing its dynamics" [7].

In a counterrotating merger, the angular momenta of the two progenitor galaxies would partially or fully cancel, leaving the remnant with little net spin. This is physically distinct from the standard slow-rotator formation pathway, which involves many successive mergers gradually erasing ordered rotation. A single counterrotating collision could, in principle, produce a dispersion-dominated system in a fraction of the time.

The team considered three formation pathways [7]:

1. Counterrotating major merger — Two galaxies spinning in opposite directions collide and cancel each other's angular momentum. Supported by the observed light asymmetry.
2. Isotropic gas infall — Rapid, chaotic gas accretion strips rotation before star formation ceases. Consistent with Magneticum simulation predictions, where kinematic transformation occurs via isotropic gas infall during starbursts [6].
3. Multiple minor mergers — Gradual accumulation of small satellite galaxies. Considered less likely given the galaxy's youth and compact structure.

How Firm Is the "No Rotation" Claim?

Any measurement of galaxy kinematics at z > 3 operates at the edge of instrumental capability, and the skeptic's position on this finding deserves serious consideration.

The most straightforward objection is geometric: if we happen to be viewing a rotating galaxy nearly face-on, the rotation axis points toward us, and the line-of-sight velocity component of the rotation vanishes. The galaxy would appear non-rotating regardless of its actual kinematics.

The authors addressed this directly. Given the galaxy's observed ellipticity of ε = 0.12 and assuming an intrinsic oblate spheroid shape, they derived an inclination angle of approximately 28–40 degrees [2]. At such an inclination, significant rotation — if present — would produce detectable velocity gradients across the galaxy. The Sérsic index of n = 3.72 further indicates an elliptical morphology rather than a face-on disk [2]. Additionally, the dynamical mass derived from the observed velocity dispersion agrees with the stellar mass from spectral energy distribution fitting, which would not be the case if substantial rotational energy were being hidden by projection effects [2].

A second concern involves spectral resolution. NIRSpec's G235M grating delivers a spectral resolution of R ≈ 1000, corresponding to a velocity resolution of about 300 km/s [2]. For a galaxy with rotational velocities below this threshold, rotation could be smeared out. However, the comparison galaxy XMM-VID3-2457 — observed with identical instrumental setup — showed clear rotation with velocities up to 400 km/s [2], demonstrating that the instrument can detect rotation at these redshifts when it is present.

A third possibility: XMM-VID1-2075 could be an ongoing merger viewed at a specific moment when the velocity fields of two merging components partially cancel. The observed light asymmetry supports ongoing or recent interaction. In this reading, "does not rotate" describes a transient snapshot of a dynamic event rather than a settled kinematic state.

These caveats are real but bounded. The research team's analysis accounts for the most common sources of false-null results, and the data quality — central spaxel signal-to-noise ratio of approximately 40, with velocity uncertainties of roughly 25 km/s — is high by the standards of high-redshift IFU spectroscopy [2].

Implications for ΛCDM and the "Too Big, Too Soon" Problem

Since its first deep-field observations, JWST has repeatedly found galaxies that appear too massive, too luminous, or too mature for their cosmic epoch. Multiple galaxy candidates at z ≈ 7–10 push against the upper limits of dark matter halo masses allowed by the ΛCDM cosmological model, implying baryon-to-stellar-mass conversion efficiencies exceeding 80 percent in some cases [8][9]. This tension — sometimes called the "too big, too soon" problem — has generated hundreds of papers and considerable debate about whether the standard model requires modification.

Source: OpenAlex

Data as of Jan 1, 2026CSV

XMM-VID1-2075 adds a new layer to this discussion. Its mass alone — 3.3 × 10¹¹ solar masses at z = 3.449 — is at the upper end of what simulations produce, but not strictly outside their predictions [2]. The MAGAZ3NE survey from which this galaxy was drawn has already demonstrated that ultramassive galaxies at z > 3 formed the bulk of their stars in intense bursts at z ≈ 4–6, producing hundreds to thousands of solar masses per year before rapidly quenching [10]. These formation histories are extreme but accommodated by current models with appropriate fine-tuning of feedback prescriptions.

What the non-rotating kinematic state adds is a timescale problem distinct from the mass problem. Even if simulations can produce galaxies this massive by z ≈ 3.4, can they also make them kinematically evolved — dispersion-dominated, slow-rotating, structurally compact — within the same timeframe? The Magneticum simulation says yes, in small numbers, via isotropic gas infall during starbursts [6]. But other simulations offer different predictions. IllustrisTNG and EAGLE have been less definitive about whether slow rotators can form this early, and disagreement among models highlights genuine uncertainty in the subgrid physics — star formation, AGN feedback, gas accretion geometry — that drives kinematic evolution in simulations [2][11].

The finding does not break ΛCDM. A single object that falls within the rare tail of simulated predictions is not a statistical challenge to the model. But it does tighten the constraints: whatever combination of physical processes built XMM-VID1-2075 had to simultaneously produce enormous stellar mass, quench star formation, and erase organized rotation, all within less than two billion years. Each of these requirements individually is demanding; all three together leave little room for slow, gradual processes.

What Comes Next: Verification and Broader Surveys

Forrest and his team have been explicit that a sample of three galaxies — however well-observed — cannot tell us whether XMM-VID1-2075 is a one-in-a-million anomaly or the first confirmed member of a population that simulations already predict at low frequency [1].

Several paths forward exist. JWST itself remains the primary tool for extending this work. The NIRSpec IFU can observe more massive quiescent galaxies at z > 3, building a statistical sample of kinematic measurements. The GA-NIFS (Galaxy Assembly with NIRSpec IFU Survey) program, designed to observe 55 galaxies at z ≈ 3–11, is positioned to provide additional kinematic data, though its target selection emphasizes different galaxy types [12].

Ground-based spectroscopy offers complementary capabilities. Forrest noted that "you can do these kinds of studies from the ground, but it's very difficult to do with high-redshift galaxies because they appear much smaller in the sky" [7]. The Extremely Large Telescope (ELT), with its 39-meter primary mirror and planned HARMONI integral field spectrograph, will deliver angular resolution exceeding JWST's in the near-infrared, enabling kinematic mapping at finer spatial scales. First light for the ELT is expected around 2028 [13].

ALMA (Atacama Large Millimeter/submillimeter Array) offers another route. While XMM-VID1-2075 has negligible ongoing star formation, molecular gas tracers could reveal residual gas kinematics or constrain the merger history through cold gas morphology. The MAGAZ3NE survey has already obtained ALMA observations of some targets, finding that ultramassive quiescent galaxies at 3 < z < 4 are extremely dust-poor [14].

The Bigger Picture

Galaxy formation theory has always operated on the assumption that angular momentum is one of the most persistent properties a galaxy can have. Gas acquires spin from cosmological tidal torques as it falls into dark matter halos; that spin is largely conserved during collapse; and the resulting disk or rotating structure persists unless disrupted by major events [15]. Finding a galaxy that has already lost its rotation less than two billion years after the Big Bang suggests that some of the universe's most massive structures assembled through unusually violent and rapid processes.

Whether XMM-VID1-2075 represents a rare outlier or the tip of a hidden population remains unknown. What is clear is that JWST's ability to resolve the internal kinematics of galaxies at z > 3 has opened a new observational window. Before Webb, astronomers could measure the masses, ages, and star formation rates of distant galaxies but not how their stars moved. That changed with NIRSpec's IFU mode, and the first results — including this one — suggest that the kinematic diversity of the early universe may be broader than models predicted.

"There are some simulations that predict very small numbers of these non-rotating galaxies very early," Forrest said [1]. The question now is whether the simulations predicted the right number.