A ladder of conclusions on the history of dark energy: an inception at z ≃ 4 preferred by Planck, DESI DR2, and Pantheon+ data

We confront a family of dark-energy histories parametrized by the cosmological event horizon, ρ_de(t) = f(t)/R_E²(t), with Planck 2018 (TT, TE, EE), DESI DR2 baryon acoustic oscillations — including the 2026 Lyman-α full-shape update — and the Pantheon+ supernova Hubble diagram, using ΛCDM and the w0–wa tangent as external controls. The data build a ladder of conclusions: (1) a fit with dark energy is decisively better than one without; (2) an evolving density beats a constant by Δχ² = 4–9, above the ±3 cross-pipeline systematic; (3) dark energy that is born — absent at high redshift, activating at z_c = 4.2 (+1.0/−0.3) (z_c > 3.4 at 95%) — beats every eternal form: an eternal screen loses by +37 at equal parameter count, and the born Avrami form leads the full three-probe battery, with w0 = −0.95 to −0.96 pinned along the residual degeneracy of the fit (marginal posterior −0.96 ± 0.03) and an early tail Ω_de(z=2.33) = 2.6–4.3%. (4) The data cannot select a model of dark-energy inception within the top class: the leading fits differ by at most 3.3 units of χ², comparable to the ±3 cross-pipeline systematic, leaving open how the birth proceeded (nucleation versus global ordering) and where the component resides (a condensate on the horizon screen versus a bulk component reproducing the same density history). We record a mild preference for the horizon reading on grounds of interpretive economy, not of χ².