Long story short: retailers are increasingly using artificial intelligence not just to tick a box, but to solve specific business tasks. And yes, it's starting to bring in money.
NVIDIA conducted a survey among retail and CPG companies. The goal was to understand how deeply AI is embedded in their operations, what's actually working, and what's still stalling. The study involved over 300 executives from various countries, including the US, UK, China, India, and other markets.
What's Happening with Investments
Let's start with the numbers. About 70% of surveyed companies plan to increase AI spending in 2026. This isn't just talk – many are already spending significant amounts. Almost half of the study participants invest more than $10 million annually in AI projects.
An important point: companies aren't just experimenting. More than half of respondents stated they are already seeing measurable returns from AI implementation. This means the technology has moved from the category of «interesting but unclear» to the rank of tools that genuinely impact business metrics.
At the same time, the scale of adoption is growing. While AI used to be used sporadically – for example, for automating customer support or warehouse management – now companies are increasingly embedding it into multiple processes simultaneously.
Where AI Is Used Most Often
Where AI Is Used Most Often 🛒
The most popular use cases are areas where mistakes are costly and manual work takes up too much time.
In first place is supply chain management. Here, AI helps predict demand, optimize delivery routes, and manage inventory. Simply put, the system suggests how much product to order, when, and where to send it so there's neither a shortage nor overstocking.
The second most popular direction is customer personalization. Product recommendations, individual offers, and interface adaptation for a specific person – all this works based on analyzing previous purchases, site behavior, and preferences. The goal is simple: show the client what really interests them, not a random set of items.
Third place goes to data analysis and forecasting. Companies use AI to find patterns in large volumes of information: what sells better at a certain time of year, how demand changes by region, and which products are bought together. This helps plan the assortment and marketing campaigns.
Solutions for automating warehouses and stores are also actively developing – from robots that pick orders to computer vision systems that monitor stock on shelves or assist in the self-checkout process.
What Is Hindering Adoption
Despite the growth in investment and positive results, problems remain. And they are quite typical for any industry trying to scale AI usage.
Infrastructure. Many companies run on legacy systems that aren't designed to process large volumes of data in real-time. For AI to work effectively, servers, storage, and data networks need updating. This is expensive and time-consuming.
Data. AI learns from data, and if it is siloed, incomplete, or poorly structured, the result will reflect that. Many retailers face the issue that information is stored in different systems that are poorly integrated with each other. For example, sales data sits in one database, customer information in another, and inventory in a third. Uniting all this into a single picture is a non-trivial task.
Competencies. AI implementation requires specialists who understand both the technology and business processes. Such people are in short supply, and training takes time. Plus, interaction needs to be built between the IT department and business units, which doesn't always go smoothly.
Security and regulation. The more data a company collects, the higher the risk of leaks and the stricter the regulatory requirements. Retail companies work with personal customer information, and any mistake can be costly – both financially and reputationally.
Generative AI Enters the Game
Generative AI deserves special mention – technologies like large language models that can create text, images, and answer questions. In retail, they are starting to be used for several tasks.
For example, for creating product descriptions. Instead of writing texts manually for thousands of items, companies generate them automatically based on product characteristics. This saves time and allows new products to be launched faster.
Another option is virtual assistants that help customers on the website or in the app. They answer questions, help choose a product, and suggest where it can be picked up or how to process a return. Importantly, such systems are becoming increasingly natural in communication and can take the context of the conversation into account.
Generative models are also used for creating marketing materials – banners, social media posts, and email newsletters. The system adapts content for different audience segments, which increases campaign effectiveness.
But there is a nuance here: generative AI is currently less predictable than classic algorithms. It might produce an inaccuracy or phrase something in the wrong way. Therefore, most companies use it with human oversight – the model offers options, and a specialist selects the best one or refines it.
What's Next
Judging by the survey, retailers are set to continue investing in AI. Moreover, the focus is shifting from experiments to scaling: companies don't want to just launch a pilot project but embed the technology into all relevant processes.
At the same time, questions of integration, data quality, and team training remain open. Technology develops quickly, but organizational changes take time. And this specifically – the company's ability to restructure processes for new tools – often turns out to be the main limitation.
In any case, AI in retail no longer looks like a futuristic story. It is a working tool that helps save money, improve service, and react faster to market changes. The question now is not whether it's worth using, but how to do it effectively.