European retailers deploy Nvidia models and warehouse simulations

Retailers and technology companies across Europe are using Nvidia-powered open models and simulation tools to improve shopping services and physical operations. Shopify post-trained NVIDIA Nemotron 3.5 Lightning for a data-retrieval subagent in Sidekick, its merchant assistant. On Shopify’s internal benchmark, the model surpassed a leading frontier model in accuracy while cutting serving costs by 90% and delivering responses 7x faster.

Zooplus worked with Infosys and Nvidia technologies on a product-discovery prototype deployed in its AWS environment, achieving 86% search accuracy. Photoroom’s image-editing deployment at Vestiaire Collective was associated with a 5% higher rate of liquidity and 3% to 7% more unique views per product page. Mirakl uses fine-tuned open models to organize vendor catalogs while maintaining control over its infrastructure.

Digital twins are also supporting creative production and logistics. L’Oréal uses Omi’s product twins in generated brand environments and is integrating NVIDIA ALCHEMI into ingredient research to move from months of experimentation to hours of computational screening. Otto Group expanded its supervised autonomous-warehouse initiative to its largest fulfillment center in Haldensleben, Germany, allowing teams to design, test and optimize operations in simulation before physical implementation.

Virtual fitting provider CATCHES reported conversion up 55% at AMIRI after seasonal adjustment against comparable styles. At GTC Berlin, Nebius is showcasing early-access Nvidia retail blueprints covering store-manager assistance, inventory simulation and optimization, and workflows that continuously generate training data and fine-tune open models.