Should broadband subsidies make service affordable or build it where service does not exist? I evaluate the $14 billion demand-side Affordable Connectivity Program (ACP) and the $42 billion supply-side Broadband Equity, Access, and Deployment (BEAD) program. I combine continuous-treatment event studies with a structural model of demand, pricing, and entry spanning more than 70,000 census tracts. I assemble five federal data sources into a new national tract-level dataset covering 727,601 potential provider-tier positions. Because the number of feasible product- positioning configurations grows exponentially with potential product lines, I use non-dominance restrictions to bound fixed costs with moment inequalities rather than repeatedly solving the entry game during estimation. ACP exposure reduced posted prices by 7–10 percent at peak and shifted offerings toward faster plans, while BEAD’s effects emerge only after its construction lag, consistent with entry. A 75 percent BEAD subsidy generates roughly $7 in welfare gains per dollar of public spending over five years. For ACP’s direct affordability channel, welfare gains roughly match its fiscal cost. The gains concentrate along different margins: BEAD’s are largest in monopoly markets, while ACP’s are greatest among low-income households. The programs are therefore complements: build where deployment is unprofitable, then subsidize affordability where price binds.