We study a model of credit markets with adverse selection where ex-ante identical lenders invest in a screening technology to reduce their type I and type II error in identifying good borrowers. A rich market structure emerges in equilibrium with continuous heterogeneity in lender screening and non-assortative matching between lenders and borrowers. Furthermore, the equilibrium features a hockey stick interest rate schedule—a segmented market structure with variable degrees of fragmentation across different level of borrower opacity. We demonstrate that this market structure is robust to changes in the screening technology as well as lender entry. We then use the model to study the impact of AI adoption and mandatory data sharing regulation on the credit market and show that while AI adoption leads to more financial inclusion, data sharing does not benefit the underserved population and counterintuitively, increases the inequality in financial access.