Coupang Inc. Builds AI Factory with NVIDIA to “Refine Rocket Delivery”
Ashish Suryawanshi, Vice President of Engineering at Coupang Inc., is presenting at the NVIDIA AI Conference & Expo on March 17. (Photo: Coupang Inc. Website)
Coupang’s parent company, Coupang Inc., has built an Artificial Intelligence (AI) factory with NVIDIA.
Coupang Inc. announced the results of its collaboration with NVIDIA at the ‘NVIDIA AI Conference & Expo’ (GTC 2026) held in San Jose, USA, on March 17 (local time). The core of this collaboration is its participation as a launch partner for ‘Dynamo 1.0’, NVIDIA’s newly unveiled open-source agent inference software.
Dynamo is known in the industry as the ‘Operating System (OS) of AI’. It plays the role of maximizing AI inference performance by organically coordinating Graphics Processing Unit (GPU) resources within data centers.
Coupang expects to increase its data throughput by up to 30 times through the adoption of Dynamo. This means that demand forecasting and delivery optimization, key components of Rocket Delivery, will be made much more sophisticated and faster than before. In fact, after the introduction of the ‘Coupang Intelligent Cloud’ (CIC) system last year, the GPU utilization rate was found to have increased from the previous 65% to 95%.
Ashish Suryawanshi, Vice President of Engineering at Coupang, stated, “With this collaboration, we have prepared a ‘Paved Road’ where engineers can focus solely on model development without being tied down to building infrastructure,” adding, “We will continue to innovate to keep the promise of Rocket Delivery.”
Jung Min-hee
댓글
0