Uber Taps AWS-Designed Chips to Train AI Models, Boost Ride and Delivery Efficiency

Uber’s ride-hailing and delivery platforms rely on real-time AI models to match drivers with riders or orders, plan routes and estimate arrival times. As trip volumes grow worldwide, computing speed and cost directly affect wait times and service reliability, making the adoption of Amazon Web Services’ in-house chips a key part of Uber’s push for more efficient infrastructure.

Uber recently announced that it is expanding its use of AWS Graviton processors and Trainium AI accelerators to support its Trip Serving Zones infrastructure and train related models. The models will power real-time decisions for rides and deliveries worldwide, improving matching, routing and arrival-time forecasts. Uber and AWS have not disclosed the investment amount, deployment date or expected cost savings.