We propose a geometric framework — Quantum Granodynamics of Spacetime (QGDS) — in which spacetime is modeled as a collection of topological boundaries between two coexisting three-dimensional domains, A and B. Each boundary element is a closed 2-sphere S2, whose dynamics is governed by an effective Helfrich bending action. This action is not introduced by analogy with biological membranes but is the most general quadratic functional of the mean and Gaussian curvatures available for a closed 2-surface — a consequence of the symmetry and differentiability of the boundary geometry. Through a geodesic coarse-graining procedure, the Einstein-Hilbert action appears as the effective macroscopic limit, with Newton's constant emerging as G_Newton = 3R_G / (16π²κ_eff) and the cosmological constant as Λ_eff ∝ σ/R_G. The modal spectrum of the boundary yields an exact zero-energy mode (l = 0) consistent with homogeneous cosmic expansion and a first physical mode at l = 2, consistent with the quadrupolar character of gravitational radiation. The framework makes observational predictions that are falsifiable in principle, including a preliminary estimate of ~0.56–0.89% systematic variation in planetary orbital velocities with galactic position, potentially testable by the Roman Space Telescope (2027). Current observational constraints and open sectors are discussed explicitly.