Uber Expands AWS Partnership Using Custom Silicon for Real-Time Operations and AI

Real-Time Infrastructure Scaling

Uber has announced a significant expansion of its infrastructure and artificial intelligence capabilities on Amazon Web Services (AWS). The ride-sharing leader is now deploying AWS Graviton4 and Trainium3 chips to power its core logistics and machine learning workloads. This transition is designed to manage the platform's 40 million daily trips more efficiently while reducing overall operational costs.

Powering the Trip Serving Zones

The company is migrating its critical "Trip Serving Zones" to the ARM-based Graviton4 processors. These zones function as the real-time heart of the app, handling the millisecond-level decisions required for matching riders with drivers and calculating routes. By using custom-designed silicon, Uber can process location data and execute millions of predictions instantly even during massive demand spikes. Beyond speed, the adoption of Graviton4 helps Uber reduce its energy consumption across its global data footprint. The high-performance nature of these chips ensures that the platform remains reliable and secure while scaling across 72 different countries. This infrastructure shift is vital for maintaining the responsiveness that millions of users count on during rush hours and major events.

Accelerating AI with Trainium

Uber is also piloting AWS Trainium silicon to train the complex AI models that define the user experience. These models analyze billions of past rides and deliveries to provide accurate arrival estimates and personalized recommendations. Trainium provides a cost-effective alternative to standard GPU infrastructure, allowing for faster iterations on large-scale machine learning tasks.

Kamran Zargahi, vice president of engineering at Uber, noted that milliseconds matter at the scale the company operates. He explained that moving these workloads to AWS provides the flexibility needed to handle global demand without disruption. This technology foundation allows Uber to focus on improving service for its massive daily user base.

Rich Geraffo, vice president at AWS, emphasized that powering one of the most demanding real-time applications in the world demonstrates the strength of their silicon. He stated that the collaboration helps Uber deliver the reliability customers expect today while building the AI-powered features of tomorrow. The partnership marks a new chapter in how logistics giants utilize specialized cloud hardware.