India debates an official caste census, yet AI has already inferred and solidified caste prejudices from existing data. Rather than disrupting society, AI formalizes embedded social codes through proxies like surnames, location, and education. Credit scoring and matrimonial platforms exemplify how algorithms perpetuate bias, making decisions appear neutral while lacking accountability. These automated systems, built on caste-shaped markets, reproduce inequalities, making them harder to detect or appeal. The author critically argues that if errors are wired into India's digital infrastructure, they could harm vulnerable populations, emphasizing that AI already knows caste, and the state must catch up to what the market infers.