The S8 tension is developing a redshift profile, and that profile is itself a problem. Low-redshift probes consistently infer less matter clustering than Planck-normalized ΛCDM predicts: KiDS-1000 gives S8 = 0.759 +0.024/-0.021, DES Y3 gives 0.776 +/- 0.017, and HSC-Y3 cosmic shear sits 2 to 3 sigma below the CMB value of 0.832 +/- 0.013. The newer development is that the suppression appears concentrated at late times: cross-correlations of ACT DR6 and Planck PR4 CMB lensing with DESI Legacy galaxies (arXiv:2410.10808), eROSITA cluster counts (arXiv:2410.09499), and HSC-Y3 shear all hint that the lowest-redshift bins drive the deficit, while higher-redshift bins agree with the CMB. The growth history looks like it bends away from the ΛCDM expectation at low z, a kink rather than a uniform offset.
This shape is exactly what ΛCDM cannot produce. Once the six parameters are fixed at recombination the growth history sigma8(z) is rigidly determined; there is no freedom for the growth rate to change character in the dark-energy era. The standard patches fail on the shape: massive neutrinos suppress power at all redshifts nearly uniformly, and the sum of masses needed to explain the full offset conflicts with other constraints; early dark energy modifies the sound horizon, not late-time growth. Explicit modified-growth analyses (arXiv:2505.09176) find that models with rapid suppression at low redshift fit HSC-Y3 plus Planck plus ACT DR6 lensing better than ΛCDM, underlining that the data want a late-time, redshift-dependent effect the standard model has no dial for.
The standing is emerging rather than settled: each individual probe carries 2 to 3 sigma, but cosmic shear, CMB-lensing cross-correlations, cluster abundances, and redshift-space distortions point in the same direction independently, and the related growth-index measurement (gamma = 0.633 +/- 0.025 versus the GR value 0.55, a 3.7 sigma exclusion) belongs to the same family. Whether the kink sharpens into a discovery or dissolves into systematics will be decided by DES Y6, Euclid, and Rubin-LSST within a few years.