Horizontal Gravity

The Core Thesis

Foundation models have introduced a qualitatively new force in digital markets: horizontal gravity. Unlike classical data gravity — which pulls computation vertically toward data stores — horizontal gravity acts laterally across application boundaries, drawing disparate data types, software capabilities, and user workflows into unified AI-mediated platforms.

The result is a morphological shift from discrete application "bundles" to amorphous capability "clumps" where traditional product boundaries lose structural significance.

Three Mathematical Frameworks

Percolation Theory determines when the phase transition occurs — the critical capability threshold at which fragmented application silos must coalesce into a connected platform cluster.

The Free Energy Principle explains why integration occurs — AI platforms minimize variational free energy by absorbing data from adjacent silos, making full integration the thermodynamic ground state.

Fisher Information Geometry describes the structure of convergence — the Riemannian manifold along which all sufficiently capable models converge to a canonical representation geometry.

The Horizontal Gravity Index (HGI)

The paper introduces a composite quantitative metric: HGI(t) = α·S(t) + β·A(t) + γ·V(t), derived from SaaS consolidation rates, API integration density, and vertical market capitalization decline. Six industries are tracked through 2030:

Developer Tools and Education Technology lead the absorption curve, with projected HGI scores of 0.90 and 0.85 respectively by 2030. Healthcare IT lags due to regulatory friction (HIPAA, FDA). Financial Services accelerates post-2026 as AI agents cross the percolation threshold for complex workflows.

Key Implications

Horizontal gravity creates zero escape velocity — the condition where switching costs of leaving an AI-integrated platform exceed the value of any standalone alternative. Niche SaaS vendors face existential risk. The paper predicts enterprise SaaS applications will converge to 100–150 per organization by 2028, and occupational categories defined by single-application expertise will decline 20–40% by 2030.

Three boundary conditions constrain the force: regulatory friction (DMA/GDPR), open-source commoditization, and diseconomies of scope.