Drug developers have historically sought broad “genus” patents covering millions or billions of chemical structures designed to keep competitors far away from a protected drug. This may no longer be possible. In the United States, many cases over the past decade—including the 2023 Supreme Court decision Amgen v. Sanofi—have invalidated broad patents because they fail to disclose sufficient information about the many compounds they cover. Our analysis of a genus patents dataset indicates that large portions of claimed chemical space are far enough from any disclosed example that standard cheminformatics tools would not predict shared biological activity, suggesting that these patents claim beyond what their data credibly support. The approved drug is structurally central among disclosed examples, suggesting that broad patents function as strategic “moats” around identified leads rather than as placeholders for exploration. Narrowing patents’ scope may open space for competing innovators without chilling early-stage research.