In the past decade, governments around the world have made significant investments to improve Internet access in underserved areas. Yet there is scant evidence on how effective these measures are at increasing Internet adoption and narrowing the digital divide. In this paper, we empirically investigate the effect of the supply of Internet access on Internet adoption. The main empirical challenge is the issue of endogeneity, as Internet service providers (ISPs) selectively enter markets based on local demand characteristics. We propose a novel solution to this problem using a panel dataset of households that relocate to different Internet access supply conditions. We find that the supply of Internet access has little effect on Internet adoption: underserved markets - those with fewer than 4 ISPs - exhibit levels of adoption comparable to well-served markets with plenty of service providers. Our findings suggest that the highly touted government initiatives to boost Internet access in underserved areas are likely to have limited impact on adoption rate in those regions.