Uber expands AWS deal to run ride-matching on Amazon's AI chips, challenging cloud rivals

Uber Shifts Critical Infrastructure to AWS AI Chips

Uber announced on Tuesday that it is expanding its cloud services agreement with Amazon Web Services to run more of its ride-sharing features on Amazon's custom chips1

1

. The ride-hailing company will particularly expand its use of AWS's Graviton4, a low-power ARM-based server CPU, and start a new trial testing Trainium3, AWS's competitor to Nvidia in the AI training chip market2

2

. This cloud partnership expansion marks a strategic shift for Uber, which operates at a scale where milliseconds matter and compute cost efficiency directly impacts profitability.

The deal enables Uber to use Amazon Web Services' Graviton and Trainium chips to support smoother rides and deliveries while training AI models that power its apps2

2

. At Uber's scale, which reached more than 40 million trips a day in 2025 across 72 countries, the infrastructure demands are substantial and latency tolerance is essentially zero3

3

. Every time a rider opens the app, a system called Trip Serving Zones determines which drivers to consider, how to weight them, and how quickly to return a match, all before the user has finished watching the loading animation.

Real-Time Trip Matching Gets Performance Boost

The announcement covers two distinct workloads that showcase Uber's AI capabilities expansion. Trip Serving Zones, Uber's real-time infrastructure for matching riders and drivers, will run on Graviton4, Amazon's ARM-based processor designed for high-throughput, low-latency compute3

3

. The workload requires responsiveness under load, particularly during demand spikes when ride volumes surge and the matching system must scale without introducing delay. Kamran Zargahi, Uber's vice-president of engineering, described the operational rationale plainly: "Uber operates at a scale where milliseconds matter. Moving more Trip Serving workloads to AWS gives us the flexibility to match riders and drivers faster and handle delivery demand spikes without disruption"3

3

.

Uber is working to optimize its digital interface, accelerate ride-matching and personalize user experiences to attract users and gain a competitive edge2

2

. The company has recorded 13.567 billion trips over its lifetime and serves more than 200 million monthly active users, generating a continuous stream of behavioral data on driver allocation, estimated arrival times, demand patterns, and route optimization3

3

.

AI Model Training on Trainium3 Begins

Separately, Uber is beginning a pilot for AI model training on Trainium3 using data from its accumulated trip history3

3

. Training AI on that dataset is a longer-term initiative, but the economics of Trainium3 make the pilot financially rational. Each Trainium3 chip delivers 2.517 petaflops in MXFP8 precision, with 144 GB of HBM3e memory and 4.9 terabytes per second of memory bandwidth. At scale, Trainium3 runs at roughly 30 to 50 per cent of the cost of comparable Nvidia H100 or H200 hardware3

3

.

Amazon is investing heavily in growing the appeal of its custom chips and attracting enterprise customers to capitalize on booming demand for AI model training and inference2

2

.

Multi-Cloud Strategy Creates Competitive Leverage

The AWS deal is the third major cloud relationship Uber has entered in the past three years. While Uber historically ran its own data centers, back in 2023, the ride-hailing company signed giant, multi-year cloud computing deals with Oracle and Google Cloud1

1

. The idea was to move the majority of its infrastructure off its own datacenters and onto these two clouds. Even in December, Uber publicly reiterated that goal, writing in a blog post about transitioning from on-premise data centers to the cloud using Oracle Cloud Infrastructure and Google Cloud Platform1

1

.

This multi-cloud strategy means Uber is effectively a significant customer of all three major hyperscalers simultaneously. The practical consequence is that Uber has unusual leverage in negotiations with each provider and unusual freedom to route workloads toward whichever platform offers the best performance-cost ratio for a given function3

3

. Moving Trip Serving Zones to Graviton4 is a statement about where AWS currently sits on that curve for high-frequency, latency-sensitive infrastructure.

Amazon's Silicon Strategy Gains Momentum

Uber joins Anthropic, OpenAI, and Apple as big tech companies that have signed on or increased their usage of AWS because of these AI chips1

1

. In December, Amazon CEO Andy Jassy said Trainium was already a multibillion-dollar business. This deal is a thorough challenge by Amazon to AWS's cloud competitors, Google and Oracle. For Uber, the move represents building a technology foundation that will make every experience smarter, keeping focus on the people who use Uber every day. Rich Geraffo, vice-president and managing director for North America at AWS, framed the partnership in terms of Uber's real-time demands: "Uber is one of the most demanding real-time applications in the world, and we're proud to be an important part of the infrastructure powering their global operations"3

3

. The Trainium3 pilot is a test of whether Amazon's AI training economics can compete with the GPU-based infrastructure Uber already has access to through its existing cloud relationships, with implications for how enterprise customers evaluate alternatives to Nvidia for model inference and training at scale.