Coupang taps Nvidia to scale AI in logistics network

Coupang Inc. has established an AI Factory with Nvidia to scale AI logistics and delivery.

It integrated its Coupang Intelligent Cloud with the Nvidia DGX SuperPOD to let engineers and data teams build and test AI models.

The AI Factory is being applied to inventory management, delivery route design, and bin packing at its logistics centers.

GPU utilization rose from 65% to 95% and that product loading efficiency improved.

The partnership would support its Rocket Delivery service by enhancing AI and machine learning operations.

🔗 Source: Chosun Daily

🧠 Food for thought

Implications, context, and why it matters.

The “AI factory” runs on Coupang Intelligent Cloud, an internal cloud platform. Coupang connected it to the NVIDIA DGX SuperPOD so engineering and data groups can train and test AI models, plus use a self-service AI ecosystem across Coupang’s global organization 1.
The platform follows a “paved road” approach. It keeps Coupang data scientists and machine learning engineers focused on model work instead of day-to-day infrastructure upkeep 1.
Teams use the top layer through a command line interface (CLI), application programming interface (API), or user interface (UI). Those tools sit on Coupang Intelligent Cloud plus NVIDIA DGX SuperPOD 1.

The custom AI platform lays out a path for getting more from costly hardware, while reducing dependence on standard cloud AI services 1.
The setup can guide other sectors with large-scale physical operations. It turns AI infrastructure into a shared self-service utility that speeds experimentation across teams 1.
Coupang also backs NVIDIA as a launch partner for Dynamo, open-source software that runs “agentic” AI systems, meaning models that can take multi-step actions, during inference 1. It frames the relationship as more than a basic customer-vendor deal 1.