Uber Partners With Amazon To Use Custom Trainium And Graviton Chips For AI Model Training

Uber expands its AWS partnership to use Trainium and Graviton chips, aimed at speeding up ride-matching and optimizing AI model training for global users.

By: AXL Media

Published: Apr 8, 2026, 4:33 AM EDT

Source: Information for this report was sourced from Reuters.

Accelerating AI Training with AWS Hardware

In a strategic move to handle intensifying digital workloads, Uber has begun utilizing Amazon’s custom-designed silicon to power its artificial intelligence initiatives. The ride-hailing giant is deploying Trainium processors, which are specifically engineered by AWS to provide high-performance, cost-effective machine learning training. By shifting AI model training to this specialized hardware, Uber aims to refine the algorithms that power its core application features, from predictive pricing to route optimization.

Optimizing Delivery and Ride-Matching

The deal also includes a significant transition to Amazon’s Graviton chips. These ARM-based processors are designed to deliver superior price-performance for cloud workloads compared to traditional x86 processors. Uber is leveraging this computing power to ensure "smoother rides and deliveries," focusing specifically on accelerating the speed of its ride-matching interface. The objective is to reduce latency for users and provide a more seamless experience during peak demand periods.

Competing Through Personalization

As the ride-hailing and delivery markets become increasingly crowded, Uber is betting on AI-driven personalization to maintain its competitive edge. By using AWS custom chips for "inference"—the process of running live AI models—Uber can offer more tailored experiences to its millions of users. This includes personalized promotions, more accurate estimated times of arrival (ETAs), and optimized search results within its delivery platform.