NVIDIA Powers Major Advances in European Retail Through AI and Digital Twins

Image: NVIDIA · Source

European retailers are leveraging NVIDIA’s AI models and digital twin technology to enhance product discovery, reduce operation costs, and optimize warehouse automation, exemplified by Shopify, Zooplus, L’Oréal, Otto Group, and others.

Leading European retailers and technology firms are harnessing NVIDIA’s accelerated open AI models and physical intelligence solutions to reshape retail operations across discovery, commerce, and fulfillment. Notably, Shopify has markedly improved its AI assistant Sidekick by post-training NVIDIA’s Megatron 3.5 Lightning model, boosting accuracy while slashing inference costs by 90% and speeding up responses sevenfold. This enhancement extends beyond Sidekick, with NVIDIA’s CCluster vLLM software delivering up to 12 times lower inference costs in multiple e-commerce functions including catalog enrichment and product recommendations.

Zooplus, a top European pet retailer, revamped its product discovery system using NVIDIA’s Nemotron Retriever for semantic search and retail catalog enrichment tooling, achieving an 86% search accuracy in a prototype deployed on AWS. Similarly, Photoroom’s deployment with Vestiaire Collective standardizes around 40,000 images daily using NVIDIA GPUs, resulting in higher product visibility and increased sales velocity.

Brands leverage NVIDIA-powered 3D asset generation and digital twin workflows to scale creative output efficiently. For instance, Omi’s ProductDrop workflow, utilized by L’Oréal, integrates AI-generated images with approved digital twins to accelerate content production up to ten times faster, reducing latency substantially. L’Oréal also accelerates cosmetic ingredient discovery by integrating NVIDIA ALCHEMI and OrbMol-v2 molecular simulation into their R&D, cutting months of experimentation to computational screening completed in hours.

In warehouse automation, the Otto Group uses NVIDIA Omniverse-based simulations to design and optimize autonomous fulfillment center operations before physical deployment. ALLSIDES’ digital twin factory captures precise 3D models for retail environments, aiding inventory tracking and layout simulation for brands like adidas, Amazon, and Nike.

Additionally, CATCHES' RealFit virtual try-on system has generated significant uplifts in conversion and sales for fashion retailer AMIRI. The company’s new PredictiveFit and AI-powered shopping agent CASI utilize open NVIDIA Nemotron models to personalize customer engagement.

At GTC Berlin, Nebius introduced three NVIDIA retail blueprints now in early access, targeting employee productivity, inventory optimization through simulation, and self-improving AI workflows for continuous fine-tuning of open models. These initiatives underscore NVIDIA’s commitment to driving innovation in the European retail sector by combining open AI models, physical intelligence, and digital twin technologies to improve operational efficiency and customer experience.