The late universe is smoother than the early universe says it should be. The clustering amplitude S8, sigma_8 scaled by the square root of the matter density, is predicted by evolving Planck's CMB-calibrated initial conditions forward under ΛCDM: 0.832 +/- 0.013. The late-time measurements consistently undershoot it: KiDS-1000 cosmic shear gives 0.759 (+0.024/-0.021), DES Y3 gives 0.776 +/- 0.017, HSC-Y3 sits 2 to 3 sigma low, and thermal SZ cluster counts tell the same story through their own calibration chain. The deficit, 2 to 3.4 sigma depending on the pairing, is collectively persistent across independent instruments, wavelengths, and analysis teams: matter has visibly clustered less than the model's growth history requires.
The standard escapes are each blocked by other data. Reconciling cluster counts requires a hydrostatic mass bias of 30 to 40 percent, while weak-lensing calibrations bound it near 20: physically implausible. Massive neutrinos suppress growth, but the required sum of 0.3 to 0.6 eV is excluded by the very datasets that define the tension, and the cosmological neutrino bound has since pressed below even the oscillation floor. Modified gravity generically enhances rather than suppresses late growth. And the emerging redshift structure cuts deepest: the suppression concentrates at late times (the S8 evolution kink, the gamma = 0.633 growth-index excess), a shape no uniform mechanism produces and ΛCDM's rigid growth history cannot bend toward.
The standing is a confirmed family of related deficits, S8, growth index, lensing-is-low, SZ counts, all pointing the same direction with a common late-time, scale-linked profile, awaiting the stage-IV verdicts of Euclid, Rubin, and CMB-S4 lensing cross-correlations.