Observer-Normalized Scale Relativity: The Cosmic Microwave Background as an Inference Horizon This work is a conceptual extension of the Observer-Normalized Scale Relativity (ONSR) framework established in Observer-Normalized Scale Relativity: A Coherence Constraint on Global Physical Inference (https://doi.org/10.5281/zenodo.18356675). It applies the coherence-based inference constraints defined in that foundational work to the cosmological limit case represented by the Cosmic Microwave Background (CMB). Within the ONSR framework, observational frames are treated as locally self-consistent but intrinsically non-equivalent when compared across large separations. Global physical inference is therefore constrained by coherence rather than by the introduction of additional dynamical entities, forces, or geometric expansion of space. This paper examines the CMB under that framework and argues that it should be interpreted not as a physical boundary or edge of space, but as an inference saturation surface arising from finite signal exchange and normalization limits. The analysis distinguishes between directly observed CMB quantities—such as angular correlations and polarization spectra—and higher-order parameters that depend on global normalization assumptions. An operational coherence diagnostic is proposed, predicting that dimensionless, directly observed quantities should remain stable under independent observational comparison, while inference-dependent parameters may exhibit structured discrepancies. A clear falsification criterion is identified. This work does not introduce new physical laws, modify general relativity, or attempt empirical validation. It functions as a boundary-condition analysis for the ONSR framework, identifying the cosmological regime at which observer-normalized inference saturates. Local relativistic physics is preserved throughout. This paper depends conceptually on the definitions and inference conditions formalized in the foundational ONSR framework and the accompanying analysis of observational frame non-equivalence (On the Non-Equivalence of Observational Frames and the Necessity of Invariant Inference Constraints in Galactic Dynamics, https://doi.org/10.5281/zenodo.18444419). It does not rely on empirical results presented in subsequent disk galaxy or high-redshift analyses, but provides the global limit context in which those studies are interpreted. All subsequent empirical datasets and analyses released by the author should be understood as conditional on the inference constraints articulated in the foundational ONSR work and on the boundary interpretation developed here. Companion works are available via the author’s Zenodo profile.