Supermassive Dark Stars: Theory & Implications

• Supermassive dark stars (SMDSs) are early-universe objects primarily composed of hydrogen and helium, whose luminosity is driven by dark matter annihilation rather than nuclear fusion.
• Their cool, low-feedback atmospheres allow prolonged baryonic accretion, enabling growth to masses between 10^5 and 10^7 M☉, unlike conventional Pop III stars.
• Observational predictions include high luminosity and potential detection with JWST and Roman, with implications for seeding heavy black hole formation in the early universe.

Searching arXiv for recent and foundational papers on supermassive dark stars to ground the article and confirm relevant citations. Supermassive dark stars (SMDSs) are a proposed class of primordial stellar objects composed almost entirely of hydrogen and helium but powered primarily by dark-matter annihilation rather than nuclear fusion. In the standard dark-star picture, weakly interacting massive particles (WIMPs) concentrate in the centers of early dark-matter halos, annihilate, and deposit heat into the protostellar gas, thereby supporting a stellar configuration that is large, cool, and extremely luminous. In this framework, continued baryonic accretion can drive growth to $M_\star \gtrsim 10^5\,M_\odot$, and in some scenarios to $10^7\,M_\odot$, making SMDSs both potential precursors of heavy black-hole seeds and possible high-redshift observational targets (Freese et al., 2010).

1. Definition and physical basis

A dark star, in this usage, is not a star made mostly of dark matter. It is a star composed almost entirely of ordinary baryonic matter—primarily primordial H/He—whose dominant power source is heating from dark-matter annihilation in the stellar interior (Freese et al., 2010). The central heating law is written as

$$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$

where $n_\chi$ is the WIMP number density, $\rho_\chi$ the WIMP mass density, $m_\chi$ the WIMP mass, $\langle \sigma v\rangle_{\rm ann}$ the annihilation cross section, and $f_Q$ the fraction of annihilation energy deposited in the star (Freese et al., 2010). The fiducial particle-physics assumptions are standard thermal WIMPs with

$$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$

and

$$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$

with $10^7\,M_\odot$0 because the annihilation products are taken to be roughly $10^7\,M_\odot$1, $10^7\,M_\odot$2, and $10^7\,M_\odot$3, the neutrinos escaping while the rest thermalize (Freese et al., 2010).

Only a very small dark-matter fraction is required. The 2010 SMDS paper states that at birth the star is “mostly hydrogen, with $10^7\,M_\odot$4 being DM,” while later review and MESA-based work emphasize that the dark-matter content is typically much less than $10^7\,M_\odot$5 by mass, with one explicit $10^7\,M_\odot$6 example containing only $10^7\,M_\odot$7 of DM, or $10^7\,M_\odot$8 of the stellar mass [(Freese et al., 2010); (Rindler-Daller et al., 2014); (Freese et al., 2015)]. The point is energetic rather than gravitating: annihilation is efficient enough that a tiny admixture can dominate the luminosity.

What makes a dark star “supermassive” is the continuation of this annihilation-powered, relatively cool phase during prolonged baryonic accretion. The benchmark SMDS masses emphasized in the foundational literature are $10^7\,M_\odot$9, $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$0, and $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$1, with explicit examples at $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$2 and $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$3 [(Freese et al., 2010); (Zackrisson et al., 2010)]. These objects are predicted to be very luminous but relatively cool, with $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$4 K in the original detectability argument, radii of order AU, and luminosities around $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$5 for $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$6 stars and $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$7 for $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$8 stars (Freese et al., 2010). Later MESA calculations confirm the same broad picture, while shifting some structural details toward hotter, smaller, denser, and more luminous models in the $$\Gamma_{\rm DMHeating} = f_Q n_\chi^2 \langle \sigma v\rangle_{\rm ann} m_\chi = f_Q \rho_\chi^2 \frac{\langle \sigma v\rangle_{\rm ann}}{m_\chi},$$9 range (Rindler-Daller et al., 2014).

2. Formation conditions and growth to supermassive scale

The SMDS scenario is set in the first star-forming halos at $n_\chi$0, with host-halo masses of roughly $n_\chi$1 (Freese et al., 2010). Baryons collapse to the halo center, where the dark-matter density is already highest, and the high-$n_\chi$2 background density is enhanced by the $n_\chi$3 scaling. The canonical claim is that the first stars form at the “right place” and “right time”: halo centers in the dense early universe (Freese et al., 2010).

Three conditions are stated for dark-star formation: sufficiently high dark-matter density, trapping of annihilation products, and dark-matter heating exceeding $n_\chi$4 cooling [(Freese et al., 2010); (Ilie et al., 2023); (Freese et al., 2015)]. For a fiducial $n_\chi$5 GeV WIMP, the heating overtakes cooling at gas density

$n_\chi$6

and after further contraction to establish thermal and hydrostatic equilibrium, a dark star is said to be born at

$n_\chi$7

with an initial mass of order $n_\chi$8 (Freese et al., 2010). The review literature gives a closely related sequence, with a proto-dark-star core of $n_\chi$9 and radius $\rho_\chi$0 AU at the onset of dominant heating, followed by the formation of a hydrostatic dark star around $\rho_\chi$1 (Freese et al., 2015).

The initial fuel enhancement is provided by adiabatic contraction. As baryons fall inward, the gravitational potential steepens and pulls in more dark matter (Freese et al., 2010). A relation used in early DS work is

$\rho_\chi$2

although later MESA work notes that this fit was not used directly there; instead the Blumenthal adiabatic contraction method was applied [(Rindler-Daller et al., 2014); (Freese et al., 2015)]. Once born, the star accretes from the surrounding baryon reservoir, which in a $\rho_\chi$3 halo is quoted as $\rho_\chi$4 of baryons, corresponding to $\rho_\chi$5 of the halo mass (Freese et al., 2010).

The reason SMDSs can grow beyond ordinary Population III masses is that they remain cool. Ordinary massive Pop III stars become hot enough to emit strong ionizing radiation and shut off inflow; the 2010 conference article cites McKee & Tan’s estimate of final Pop III masses around $\rho_\chi$6 (Freese et al., 2010). By contrast, a DM-powered dark star remains cool enough that radiative feedback does not halt accretion, and so can continue growing to $\rho_\chi$7 and, in the most extreme modeled case, $\rho_\chi$8 (Freese et al., 2010). This suggests that the defining role of dark-matter heating is not merely extra luminosity, but structural regulation of the protostellar trajectory.

3. Fueling mechanisms and stellar structure

Two sustaining mechanisms recur throughout the SMDS literature. The first is extended adiabatic contraction. Earlier spherical-halo models found that the DM enclosed in the star would be consumed in $\rho_\chi$9 yr, but the SMDS papers argue that realistic halos are generally triaxial and support box and chaotic orbits that pass arbitrarily close to the center, continually repopulating the cusp with low-angular-momentum DM (Freese et al., 2010). In that picture, the central dark-matter cusp is not depleted as rapidly as in a spherical model, allowing larger and brighter dark stars and extending their lifetime from $m_\chi$0 yr to millions to billions of years in optimistic summaries [(Freese et al., 2010); (Freese et al., 2015)]. A major caveat, already emphasized in early observational critiques, is that this replenishment mechanism is uncertain and may be physically difficult to realize at the level required for $m_\chi$1 objects (Zackrisson et al., 2010).

The second mechanism is capture. Ambient halo WIMPs can scatter elastically off nuclei in the star, lose enough energy to become bound, and then annihilate inside the stellar interior (Freese et al., 2010). Capture depends on the ambient DM density and the WIMP–nucleus scattering cross section, which the conference paper treats as effectively a free parameter bounded only by direct-detection limits (Freese et al., 2010). In later stability work, captured WIMPs are modeled with the standard thermalized Gaussian profile

$m_\chi$2

and for canonical WIMP masses the resulting $m_\chi$3 is only about $m_\chi$4 of the stellar core radius, making the captured component too centrally concentrated for its gravity alone to affect the GR instability significantly (Haemmerlé, 2024). The implication in that analysis is that realistic capture alters the star mainly through heating, not through direct gravitational stabilization.

The stellar structure itself has been modeled first with polytropic approximations and later with MESA. The essential equilibrium relations are hydrostatic balance,

$m_\chi$5

and thermal balance,

$m_\chi$6

with

$m_\chi$7

in the review notation (Freese et al., 2015). Early work interpreted the stars as “puffy,” low-density, extended objects, which naturally explains how they can be simultaneously luminous and relatively cool (Freese et al., 2010). MESA calculations later found “remarkably good overall agreement” with the earlier polytropic models, while obtaining stars hotter by a factor of $m_\chi$8, smaller in radius by a factor of $m_\chi$9, denser by a factor of $\langle \sigma v\rangle_{\rm ann}$0, and more luminous by a factor of $\langle \sigma v\rangle_{\rm ann}$1 in the $\langle \sigma v\rangle_{\rm ann}$2 range relative to Freese et al. (2010) (Rindler-Daller et al., 2014). Those models also confirm that supermassive dark stars are very well approximated by $\langle \sigma v\rangle_{\rm ann}$3 polytropes (Rindler-Daller et al., 2014).

Representative MESA values for $\langle \sigma v\rangle_{\rm ann}$4 GeV illustrate the supermassive regime. In the large-minihalo case ($\langle \sigma v\rangle_{\rm ann}$5, $\langle \sigma v\rangle_{\rm ann}$6 host halo), the $\langle \sigma v\rangle_{\rm ann}$7 model reaches

$\langle \sigma v\rangle_{\rm ann}$8

with central density $\langle \sigma v\rangle_{\rm ann}$9 and central temperature $f_Q$0 (Rindler-Daller et al., 2014). These figures are still within the general dark-star characterization of huge, luminous, comparatively cool objects, although they are somewhat hotter than the earliest blackbody-based detectability estimates (Rindler-Daller et al., 2014).

4. Stability, collapse, and black-hole seed formation

The original SMDS scenario states that once the DM fuel runs out, the star undergoes a short fusion phase and ultimately collapses to a black hole (Freese et al., 2010). This outcome is one of the main astrophysical motivations of the model, because the resulting black holes are already very massive and hence plausible seeds for the $f_Q$1 black holes observed at $f_Q$2 and at galactic centers [(Freese et al., 2010); (Freese et al., 2015)]. Review treatments state more generally that remnants are expected to exceed $f_Q$3, with the largest SMDSs producing $f_Q$4 black holes (Freese et al., 2015).

Later work sharpened the gravitational-instability aspect. The paper "General-relativistic instability in rapidly accreting supermassive stars in the presence of dark matter" (Haemmerlé, 2024) does not construct canonical dark stars from first principles, but it analyzes the stability boundary of rapidly accreting SMSs in a regime that becomes SMDS-like when WIMP annihilation maintains $f_Q$5 K rather than the usual $f_Q$6 K. Its main conclusion is that dark-matter gravity alone is generally unimportant in realistic WIMP-capture configurations, because the DM distribution is too centrally concentrated, but annihilation heating can materially delay the GR instability by keeping the star puffier and less centrally dense (Haemmerlé, 2024). Quantitatively, as long as WIMP annihilation maintains

$f_Q$7

the GR instability is reached only for stellar masses

$f_Q$8

typically an order of magnitude above the ordinary H-burning case (Haemmerlé, 2024). This supports a central inference of the SMDS literature: annihilation support delays contraction and thereby delays GR collapse.

The 2025 paper "Early Formation of Supermassive Black Holes via Dark Star Gravitational Instability" (Freese et al., 11 Nov 2025) goes further by calculating collapse masses of accreting dark stars with MESA. Its central claim is that dark stars can grow by accretion to masses in the range

$f_Q$9

before the general-relativistic Feynman–Chandrasekhar instability causes dynamical collapse to black holes (Freese et al., 11 Nov 2025). The stars remain cool, diffuse, and radiation dominated, close to $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$0 polytropes, with effective temperatures $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$1 K and often maxima of $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$2 K in the simulations (Freese et al., 11 Nov 2025). The collapse threshold is discussed in terms of

$$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$3

and a critical central density

$$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$4

for primordial composition (Freese et al., 11 Nov 2025). Over most of parameter space, the collapse mass is fit by

$$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$5

yielding, for example, $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$6 and $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$7 for $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$8 GeV (Freese et al., 11 Nov 2025). This suggests a more direct route from an annihilation-powered stellar phase to heavy BH seeds than the earlier “short fusion phase then collapse” formulation.

Not all recent work supports the same stabilization mechanism. The paper "Dark Matter and General Relativistic Instability in Supermassive Stars" (Kehrer et al., 2024) studies ordinary SMSs containing a dynamically significant amount of collisionless dark matter and shows that a DM content of order $$m_\chi \sim 1\,{\rm GeV} - 10\,{\rm TeV}, \qquad m_\chi = 100\,{\rm GeV}\ \text{canonical},$$9 by mass throughout the star can raise the critical central density for GR instability, in some cases by orders of magnitude (Kehrer et al., 2024). However, that paper is explicit that it is not a model of annihilation-powered dark stars; it addresses gravitational stabilization by embedded, non-annihilating DM, and the required densities are extreme, about 20 orders of magnitude above the expected cosmological background in one $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$0 example (Kehrer et al., 2024). The distinction between DM-powered SMDSs and SMSs merely containing DM is therefore conceptually important.

5. Observational signatures and constraints

The original detectability argument rests on redshifted stellar spectra compared with instrument sensitivities. The 2010 conference paper models blackbody spectra for benchmark stars formed at $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$1 and observed at $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$2, $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$3, and $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$4 (Freese et al., 2010). For JWST the relevant bands are NIRCam at $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$5 and $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$6, with approximate resolving power $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$7, and MIRI at $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$8, $$\langle \sigma v\rangle_{\rm ann} = 3\times 10^{-26}\ {\rm cm^3\,s^{-1}},$$9, with sensitivity curves shown for $10^7\,M_\odot$00 and exposures of $10^7\,M_\odot$01 s and $10^7\,M_\odot$02 s (Freese et al., 2010). The $10^7\,M_\odot$03 SMDS would be detectable by JWST/NIRCam in a million-second exposure only if it survives down to about $10^7\,M_\odot$04, whereas the $10^7\,M_\odot$05 SMDS would be detectable even at $10^7\,M_\odot$06 in a $10^7\,M_\odot$07 s exposure in both NIRCam bands and might be marginally detectable with MIRI at $10^7\,M_\odot$08 in $10^7\,M_\odot$09 s (Freese et al., 2010). The detectability argument thus hinges mainly on mass and survival time.

The intergalactic medium is central to the observational picture. The early SMDS figures were not corrected for Ly$10^7\,M_\odot$10 absorption by the IGM, but they explicitly marked the redshifted rest-frame $10^7\,M_\odot$11 Å break; flux blueward of that wavelength is expected to be absorbed to some extent by the IGM (Freese et al., 2010). This naturally leads to dropout selection. In the HST discussion, the 2010 conference article notes that a $10^7\,M_\odot$12 SMDS formed at $10^7\,M_\odot$13 with $10^7\,M_\odot$14 GeV could already be visible in HST/WFC3-IR data as a $10^7\,M_\odot$15-band dropout (Freese et al., 2010).

That possibility prompted immediate observational tests. "Observational constraints on supermassive dark stars" (Zackrisson et al., 2010) argues that some of the proposed $10^7\,M_\odot$16 objects should already have been detectable with HST and 8–10 m class telescopes. Using TLUSTY atmosphere models over $10^7\,M_\odot$17, AB magnitudes, and complete IGM absorption below rest-frame $10^7\,M_\odot$18, it concludes that $10^7\,M_\odot$19 SMDSs at $10^7\,M_\odot$20 would be bright enough in $10^7\,M_\odot$21 to appear as $10^7\,M_\odot$22-dropouts over roughly $10^7\,M_\odot$23, yet no credible such dropouts were found in the relevant Bouwens et al. data (Zackrisson et al., 2010). The resulting upper limit on the halo occupation fraction is written as

$10^7\,M_\odot$24

where $10^7\,M_\odot$25 is the halo formation rate per unit redshift and arcmin$10^7\,M_\odot$26, $10^7\,M_\odot$27 the survey area, $10^7\,M_\odot$28 the cosmic age interval per unit redshift, and $10^7\,M_\odot$29 the SMDS lifetime (Zackrisson et al., 2010). For $10^7\,M_\odot$30 SMDSs at $10^7\,M_\odot$31, the null detection implies

$10^7\,M_\odot$32

for $10^7\,M_\odot$33 yr (Zackrisson et al., 2010). The qualitative conclusion is that such stars must be “exceedingly rare or short-lived” (Zackrisson et al., 2010).

A related 2011 study, "The observational signatures of high-redshift dark stars" (Zackrisson, 2011), also uses TLUSTY and argues that $10^7\,M_\odot$34 SMDSs should be bright enough at $10^7\,M_\odot$35 to be detectable even with HST, while lower-mass dark stars below $10^7\,M_\odot$36 remain too faint without strong gravitational lensing (Zackrisson, 2011). It further notes that if $10^7\,M_\odot$37 K, a dark star can photoionize surrounding gas and form an H II region, and Cloudy calculations under idealized assumptions can boost the H-band flux by $10^7\,M_\odot$38 mag at $10^7\,M_\odot$39 and $10^7\,M_\odot$40 mag at lower redshift (Zackrisson, 2011). This is explicitly presented as an upper limit because the nebula could be lost, extended, or depleted of gas (Zackrisson, 2011).

Subsequent JWST- and Roman-era work shifted attention to $10^7\,M_\odot$41 objects at $10^7\,M_\odot$42. "Observing Dark Stars with JWST" (Ilie et al., 2011) uses TLUSTY atmospheres, HST abundance bounds, and JWST dropout selections to argue that $10^7\,M_\odot$43 SMDSs are bright enough to be detected in all NIRCam bands, appearing as J-band, H-band, or K-band dropouts at $10^7\,M_\odot$44, respectively (Ilie et al., 2011). For a total survey area of $10^7\,M_\odot$45, the raw count estimate for $10^7\,M_\odot$46 SMDSs found as H or K-band dropouts is $10^7\,M_\odot$47, but once HST non-detections at $10^7\,M_\odot$48 are imposed, the realistic observable number is reduced to $10^7\,M_\odot$49 (Ilie et al., 2011).

More recently, Roman forecasts have extended the same observational logic. "Detectability of Supermassive Dark Stars with the Roman Space Telescope" (Zhang et al., 2023) argues that Roman can detect $10^7\,M_\odot$50 SMDSs up to about $10^7\,M_\odot$51 and, with $10^7\,M_\odot$52 lensing and $10^7\,M_\odot$53 s exposures, SMDSs as small as $10^7\,M_\odot$54 at $10^7\,M_\odot$55 (Zhang et al., 2023). That study also emphasizes a practical Roman+JWST synergy: Roman can find wide-field high-$10^7\,M_\odot$56 dropouts, but confirmation requires JWST spectroscopy, especially the proposed He II $10^7\,M_\odot$57 absorption “smoking gun” for SMDSs (Zhang et al., 2023).

6. Candidate interpretations, diffuse constraints, and current status

JWST has produced direct candidate-based SMDS interpretations. "Supermassive Dark Star candidates seen by JWST?" (Ilie et al., 2023) argues that three spectroscopically confirmed JADES sources—JADES-GS-z13-0, JADES-GS-z12-0, and JADES-GS-z11-0—are consistent with SMDS atmosphere models and therefore represent the first dark-star candidates (Ilie et al., 2023). The fitting uses TLUSTY spectra based on prior dark-star structure models, with $10^7\,M_\odot$58 GeV, $10^7\,M_\odot$59, a $10^7\,M_\odot$60 formation halo, and $10^7\,M_\odot$61 (Ilie et al., 2023). Best-fit cases include $10^7\,M_\odot$62 and $10^7\,M_\odot$63 models, with statistically acceptable $10^7\,M_\odot$64 values, though the authors explicitly note that galaxy fits are sometimes better and that the evidence is “suggestive, not definitive” (Ilie et al., 2023). The key proposed diagnostics are unresolved point-source morphology and He II $10^7\,M_\odot$65 absorption, which they treat as characteristic of a hot SMDS atmosphere, whereas a Pop III galaxy would instead show nebular emission (Ilie et al., 2023).

A 2025 follow-up, "Spectroscopic Supermassive Dark Star candidates" (Ilie et al., 9 May 2025), argues that JADES-GS-z11-0 and JADES-GS-z13-0 remain spectroscopically consistent with SMDS interpretations and adds JADES-GS-z14-0 and JADES-GS-z14-1 as new candidates (Ilie et al., 9 May 2025). It reports a tentative He II $10^7\,M_\odot$66 absorption feature in JADES-GS-z14-0 with $10^7\,M_\odot$67 and equivalent width $10^7\,M_\odot$68 Å, while also acknowledging that a probable ALMA [O III] $10^7\,M_\odot$69 emission line in the same source is in tension with the simplest isolated pristine dark-star model (Ilie et al., 9 May 2025). The paper is therefore explicit that the evidence remains circumstantial and that some candidates may require theoretical refinements involving metal-rich or non-isolated environments (Ilie et al., 9 May 2025).

At the population level, SMDS scenarios are now also being constrained by diffuse backgrounds. "Diffuse Neutrino Signals from Dark Stars Seeding Super-Massive Black Holes" (Schwemberger et al., 2024) develops the idea that a cosmological SMDS population should generate a diffuse neutrino background observable in existing experiments and relevant to SMBH seeding. It normalizes the DS fraction to an SMBH-seed requirement and obtains

$10^7\,M_\odot$70

at $10^7\,M_\odot$71, adopting a plausible range $10^7\,M_\odot$72 (Schwemberger et al., 2024). A later multimessenger analysis, "Multimessenger Constraints on Supermassive Dark Stars and Their Black Hole Remnants" (Manno et al., 3 Dec 2025), computes the diffuse electromagnetic emission from an SMDS population and its remnant BH spikes, concluding that DS-related contributions can exceed the Fermi-LAT extragalactic $10^7\,M_\odot$73-ray background for thermal relic annihilation cross sections and DM masses below $10^7\,M_\odot$74 TeV (Manno et al., 3 Dec 2025). In that work the fiducial occupation fraction is

$10^7\,M_\odot$75

chosen to be large enough to seed the SMBH population, and the strongest pressure on the model comes not from the thermal stellar emission but from annihilation outside the star and around the remnant BHs (Manno et al., 3 Dec 2025). This suggests that the observational status of SMDSs is no longer determined solely by direct imaging, but increasingly also by population-integrated diffuse signatures.

The model has also entered gravitational-wave discussions. "Reconstructing PTA measurements via early seeding of supermassive black holes" (Ghodla et al., 8 Jul 2025) treats collapse of SMDSs as a heavy-seed channel for SMBHs and argues that SMDS-seeded SMBHs with comoving seed number density of $10^7\,M_\odot$76 can be the dominant contributor to the PTA signal, with $10^7\,M_\odot$77 binaries acquiring a peak merger rate at $10^7\,M_\odot$78 (Ghodla et al., 8 Jul 2025). That paper is not about SMDS structure, but it underscores the continuing relevance of SMDSs as a heavy-seed hypothesis in broader early-universe astrophysics.

A final recent extension concerns post-collapse remnants. "JWST's Little Red Dots as collapsed Supermassive Dark Stars" (Ilie, 1 Jun 2026) argues that SMDSs, powered by DM annihilation rather than nuclear burning, naturally satisfy the structural and energetic requirements for quasi-star formation while relaxing the restrictive conditions of canonical SMS pathways (Ilie, 1 Jun 2026). In that paper a fiducial non-rotating, zero-metallicity SMDS powered by 100 GeV WIMPs reaches $10^7\,M_\odot$79, $10^7\,M_\odot$80, $10^7\,M_\odot$81, and $10^7\,M_\odot$82 at GR onset, then collapses to a prompt BH of order $10^7\,M_\odot$83 while retaining a massive envelope that can inflate into a cool quasi-star-like photosphere with $10^7\,M_\odot$84 (Ilie, 1 Jun 2026). This is not part of the canonical SMDS literature of 2010–2015, but it illustrates how the dark-star pathway has been reinterpreted as a route to later, BH-embedded luminous transients.

Overall, the SMDS concept remains a technically developed but observationally unsettled proposal. The foundational papers establish a coherent physical sequence—adiabatic contraction and possibly capture, cool annihilation-supported growth to $10^7\,M_\odot$85, and collapse to heavy black-hole seeds [(Freese et al., 2010); (Freese et al., 2015)]. Later stellar-evolution and instability work supports the basic inference that annihilation support delays contraction and permits larger collapse masses than ordinary SMS evolution [(Rindler-Daller et al., 2014); (Haemmerlé, 2024); (Freese et al., 11 Nov 2025)]. At the same time, direct-imaging constraints strongly challenge the most massive, bright, long-lived $10^7\,M_\odot$86 variants, requiring them to be very rare or short-lived [(Zackrisson et al., 2010); (Zackrisson, 2011)]. Current JWST candidate claims and diffuse-background analyses indicate that the subject has shifted from pure theory to a regime of active observational testing, but not yet to secure identification (Ilie et al., 2023, Ilie et al., 9 May 2025, Manno et al., 3 Dec 2025).