Should broadband subsidies make service cheaper, or build it where none exists? I evaluate the two largest programs of the 2021 Infrastructure Act in one welfare framework: the $14.2 billion Affordable Connectivity Program (ACP) and the $42.45 billion Broadband Equity, Access, and Deployment (BEAD) program. I pair cross-state event studies with a tract-level structural model of demand, pricing, and entry estimated on 170,419 products across 66,454 census tracts. Because the entry game is intractable at this scale, I bound provider-tier fixed costs with a moment-inequality estimator that never computes an equilibrium. ACP cut prices 7–10 percent at peak and shifted demand toward faster plans; BEAD’s effects are smaller and lagged, tracking construction timelines. In counterfactuals over a 25-year horizon, BEAD returns at least $12 in welfare per subsidy dollar under my most conservative assumptions and about $26 in the baseline; ACP’s measured gains roughly offset its outlays, a floor that omits the equilibrium price cuts the event study documents. The programs complement rather than substitute: BEAD’s gains fall on unserved and monopoly tracts and stay flat across income, while ACP’s fall on low-income households and stay flat across market structure. The sequencing rule follows: build where deployment is uneconomic, then target affordability where price still binds.