Coupang Deploys NVIDIA AI Infrastructure for Logistics

SEATTLE, WA — Coupang (NYSE: CPNG) has implemented new artificial intelligence infrastructure in partnership with NVIDIA to expand its internal machine learning capabilities.

The integration incorporates the Coupang Intelligent Cloud (CIC), an internal platform introduced in July 2025, together with NVIDIA DGX SuperPOD hardware.

The combined architecture provides a centralized environment for data scientists and engineers in Seattle and Mountain View, California, to develop and test predictive models, according to the company.

The implementation of the CIC has increased graphics processing unit (GPU) utilization from 65% to 95%, the company reported.

Coupang applies these AI models to logistical operations, such as fulfillment center inventory management, personnel scheduling, bin packing, and delivery route optimization.

Coupang is also serving as a launch partner for Dynamo, NVIDIA's open-source software designed for agentic inference, according to the company.

At the NVIDIA AI Conference & Expo on March 17, Coupang Vice President of Engineering Ashish Suryavanshi said the collaboration enables the company’s engineers to build on existing machine learning frameworks that support its operations.

The CIC platform operates across three architectural layers.

Developers access the top layer through a command-line interface (CLI), an application programming interface (API), and a user interface (UI), which abstract the underlying hardware infrastructure during model development.