Shopify Cuts AI Serving Costs 90% as NVIDIA Details European Retail Deployments

NVIDIA detailed how European retailers use its open models and physical AI tools, in a blog post published around GTC Berlin. The biggest number came from Shopify: it post-trained NVIDIA's Nemotron 3.5 Lightning for a GraphQL subagent in Sidekick, its assistant for merchants, and says the model beat a leading frontier model on its internal benchmark while cutting serving costs by 90% and answering 7 times faster.

Online pet retailer Zooplus rebuilt its product search with Infosys and reached 86% search accuracy in a prototype running in its AWS environment. Photoroom standardizes about 40,000 listing images a day for Vestiaire Collective; those listings sold at a 5% higher rate. Mirakl's Catalog Transformer, which runs open models on its own infrastructure and European clouds such as Scaleway, has onboarded more than 70 million products.

On the physical side, Otto Group extended its simulated, human-supervised autonomous warehouse project, built with Reply on NVIDIA Omniverse, to its largest fulfillment center in Haldensleben, Germany. L'Oréal is using NVIDIA ALCHEMI to screen ingredient combinations in hours instead of months, and virtual try-on company CATCHES reported conversion up 55% in its deployment with AMIRI.