The scalar spectral index n_s, the tilt of the primordial power spectrum, is the single most consequential parameter for selecting among early-universe models, and the two flagship CMB experiments now disagree about it. Planck 2018 measured n_s = 0.9649 +/- 0.0042. The Atacama Cosmology Telescope DR6 release measures n_s = 0.9743 +/- 0.0034, just over 2 sigma higher, and combinations of ACT with Planck and DESI BAO push the preferred value to 0.974 +/- 0.003, with DESI DR2 updates reaching 0.977 +/- 0.003. Both instruments observe the same sky through overlapping multipole ranges, so a shift this size in a six-parameter model's cornerstone is either an unidentified systematic in one of the world's two most scrutinized datasets or a sign that the model is bending under strain.
The consequences propagate through the entire inflationary landscape, because in ΛCDM the spectral index is a free input whose value is set by the shape of an unknown inflaton potential. At Planck's value, the Starobinsky model and alpha-attractors sat in the sweet spot; at ACT's value they are disfavored, curvaton and nonminimally coupled scenarios move in, and analyses have even revived discussion of an exact Harrison-Zeldovich n_s = 1 spectrum (arXiv:2210.09018). A parameter with no derivation can absorb any shift, but everything built on top of it reorders each time the number moves, which is precisely what has happened across the post-ACT literature.
The standing is unresolved: SPT-3G data sit between the two, the shift correlates with how each experiment weights the damping tail and with calibration and foreground choices, and no identified systematic accounts for it. CMB-S4 and further SPT/ACT analysis will settle the number; what is already clear is that ΛCDM provides no physical expectation for where n_s should land, so the disagreement can only be adjudicated by measurement, never by the model.