News
NVIDIA says European retailers are using its models, robotics and digital-twin tools for commerce, catalog enrichment, warehouse simulation and virtual try-on.
NVIDIA used a Europe retail-focused blog post on Oct. 22, 2026, to highlight how retailers and software companies are applying open models, robotics simulation and digital twins across commerce, warehouse planning and product discovery. The company described a mix of reported production deployments and prototypes, but the examples are vendor-reported rather than independently audited sector data.
Evidence and context
In the post, NVIDIA said Shopify post-trained its Nemotron 3.5 Lightning model for a GraphQL subagent inside Sidekick, Shopify’s AI assistant for merchants. According to Shopify’s internal benchmark cited by NVIDIA, the post-trained model surpassed a leading frontier model in accuracy while cutting serving costs by 90% and delivering responses 7x faster. NVIDIA also said its CCluster vLLM inference optimization software helped Shopify achieve 4.5x faster performance and up to 12x lower inference cost across five additional use cases: catalog enrichment, merchant assistant, e-commerce shopping simulation, product taxonomy classification and merchant and buyer recommendations.
The post said Zooplus, working with Infosys and NVIDIA technologies, rebuilt its product discovery stack for the agentic era. NVIDIA said a production-ready prototype inside Zooplus’ Amazon Web Services environment achieved 86% search accuracy using the NVIDIA Nemotron Retriever library for semantic search and the NVIDIA Blueprint for retail catalog enrichment for generated, search-ready product attributes.
NVIDIA also pointed to Photoroom’s deployment with Vestiaire Collective, saying the company’s Image Editing API standardizes about 40,000 primary images a day across 70 countries. The post said listings processed by Photoroom sold at a 5% higher rate of liquidity, with 3% to 7% more unique views per product page, and that the pipeline runs on NVIDIA GPUs with TensorRT and Triton Inference Server. NVIDIA said Dynamo Snapshot can help respond to demand spikes.
Other examples included Omi’s ProductDrop workflow, which the company said anchors generated images to approved digital twins, and L’Oréal’s use of Omi digital twins within its CreAItech program. NVIDIA said Omi runs TRELLIS 3D asset generation on RTX PRO 6000 Blackwell Server Edition GPUs, speeding inference by 10x and reducing pipeline latency by 4x. NVIDIA also said L’Oréal is using NVIDIA ALCHEMI to speed ingredient screening from months to hours, and is testing CurieOS with Orbital Industries’ OrbMol-v2, which NVIDIA described as five times faster than comparable models with 21% lower error than the next-best model of its kind.
The post also described Mirakl’s use of open models on NVIDIA GPUs on its own infrastructure and through European cloud providers such as Scaleway, and said the company’s Catalog Transformer has onboarded more than 70 million products.
Physical AI and retail operations
NVIDIA said physical AI and digital twins are being used to simulate retail environments, automate warehouse operations and support virtual try-on tools. It cited Otto Group’s autonomous warehouse under human supervision initiative, built with Reply on NVIDIA Omniverse libraries and running on Google Cloud, now extended to the group’s largest fulfillment center in Haldensleben, Germany. The company said the setup lets logistics teams design, test and optimize warehouse operations in simulation before physical implementation.
In another example, NVIDIA said ALLSIDES launched a 3D Digital Twin Factory that turns physical objects into SimReady 3D digital twins in minutes, with sub-millimeter geometric accuracy, measured physical properties and relightable materials. NVIDIA said those digital twins are used to simulate store layouts and track products through the shopping journey and at checkout. The post said the platform is built on CUDA, OptiX and TensorRT and already serves adidas, Amazon, Nike, Zalando and Zara.
NVIDIA also said CATCHES reported results from its RealFit deployment with AMIRI, with conversion up 55%, add-to-cart up 80% and gross sales up 56% after seasonal adjustment against like-for-like styles. It added that CATCHES launched PredictiveFit and CASI, a personal fashion agent that shops, styles and sizes for customers over WhatsApp, powered by open weight models including NVIDIA Nemotron 3 and built on NVIDIA Omniverse and the Newton Physics Engine.
What is new and what remains limited
The blog post is best read as a company-curated overview of retail use cases rather than a market-wide measurement of adoption. Several figures are tied to internal benchmarks, vendor deployments or prototype environments, and the source does not provide independent verification of the outcomes. NVIDIA’s role in the examples is mainly as an infrastructure and software platform supplier, not as a retailer.
As background, the verified recent development most directly tied to this theme is Nebius’ Oct. 21, 2026 announcement of early access to three NVIDIA retail blueprints for inventory simulation, store operations and agentic fine-tuning. NVIDIA said those blueprints were available in early access; it did not describe them as completed, large-scale European retail deployments.
Be the first to comment