For years, global supply chains were designed with one primary objective: to maximise efficiency and reduce costs, concentrating production in a few hubs and optimising transport and inventory. Today, however, this model is showing all its limitations. Geopolitical tensions, extreme weather events, energy price volatility and new regulations are making supply chains increasingly vulnerable to disruptions and unforeseen events. In this context, artificial intelligence can help companies better manage their supply chains and respond swiftly to change.

From data analysis to forecasting: what AI can already do

The applications are numerous. Algorithms make it possible to analyse large volumes of data from suppliers, warehouses, logistics networks and end markets in order to forecast demand, optimise inventory levels, plan deliveries and identify potential bottlenecks in advance. They thus become predictive tools, capable of simulating alternative scenarios and reorganising production and distribution flows in near real time. According to various analyses, this capacity for forecasting and coordination is set to become one of the key drivers of industrial competitiveness in the years ahead.

At the same time, the adoption of artificial intelligence is contributing to a broader transformation of the very architecture of supply chains. Many companies are moving away from highly centralised models based on a few large production hubs, opting instead for more distributed, regional networks. AI makes it possible to manage more complex systems — with multiple suppliers and production facilities located closer to end markets — by enabling the dynamic coordination of production, logistics and procurement. This is not a return to protectionism or the end of globalisation, but rather a reconfiguration of it: supply chains remain global, but with a more regional, flexible and data-driven organisation.

Integrating artificial intelligence into supply chains also carries risks

Up to this point, the integration of artificial intelligence might seem like a linear, almost inevitable trajectory. But things are not quite so straightforward, as an article by Agenda Digitale effectively explains. In digital supply chains, artificial intelligence is not used solely by the lead company, but by many different actors — suppliers, logistics operators, software platforms and cloud services — all of whom integrate algorithms into their own management and production systems. The problem is that AI solutions are often developed and managed by external technology vendors, while the companies using them have no full visibility into the data used to train the models or into how the algorithms actually work.

This creates new vulnerabilities. Anyone who has worked with generative AI tools knows all too well that they are prone to bias: if the data used to train the models is incomplete, distorted or insufficiently representative, the algorithms can generate systematically skewed forecasts and decisions — for instance in supplier selection, inventory management or logistical risk assessment — amplifying errors that then ripple through the entire supply chain.

Cyberattacks, too, can spread rapidly. Supply chain attacks, for example, exploit precisely the interdependencies between suppliers and digital platforms to indirectly target companies that are not the primary objective. In an ecosystem where information systems, logistics platforms and management software are in constant dialogue with one another, pinpointing where an error or manipulation has originated becomes far more complex. For this reason, alongside the benefits, growing attention is being paid to the risks associated with cybersecurity, data quality and traceability, algorithmic transparency and the governance of technologies used across the entire supply chain.