The trading floor is getting quieter, and smarter. Wall Street banks are expanding the use of machine-learning models to price assets, route orders, and manage risk, shifting work once done by rooms full of traders to a handful of engineers and their systems.

The fictional bank Meridian Securities says its automated desks now handle the bulk of routine flow, freeing human traders for complex, relationship-driven deals. Executives frame the change as evolution, not replacement — though the headcount tells a more complicated story.

Regulators are watching warily, concerned about models that behave unpredictably in a crisis. For the young analysts who once flooded onto trading floors, the message is clear: the path to a Wall Street career now runs through code.