The late Integrated Sachs-Wolfe effect is dark energy's direct signature on the CMB: as accelerated expansion stretches large-scale gravitational potentials, photons crossing them pick up a slight net energy shift, producing a positive cross-correlation between CMB temperature maps and foreground large-scale structure (Crittenden and Turok 1996). The signal is weak by construction, detectable only statistically, and the measurements keep coming in light: many galaxy- and void-based cross-correlations find amplitudes below or only marginally consistent with the ΛCDM expectation, with uncomfortable scatter between tracers and pipelines (Fang et al. 2019; Manzotti and Dodelson 2014).
The deficit is awkward because the prediction has so few moving parts. Constant Lambda fixes a uniform rate of potential decay; convolve that with the observed matter distribution and the cross-correlation amplitude follows. A persistently low signal therefore demands either lower sigma_8, which deepens the S8 tension, or modified gravity, or survey systematics subtle enough to bias many independent analyses the same way. The sector also carries an internal contradiction: while the mean signal runs low, stacked supervoids return ISW imprints several times stronger than ΛCDM allows (Granett et al. 2008), so the same effect is simultaneously too weak on average and too strong in voids.
The standing is a decade of marginal detections pointing the same direction. Euclid, LSST, and CMB-S4 cross-correlations will pin the amplitude to a few percent and, critically, will be deep enough to split the signal by sightline environment, the test that separates a statistical fluke from structure in the potential-decay rate itself.