2025 was an intense year. The technology ecosystem’s attention focused primarily on Agentic AI, a generation of systems capable of making autonomous decisions and performing actions with little or no human supervision. Unlike the deterministic models that dominated automation for years, agentic AI introduced a paradigm based on goals, context, and continuous interaction with the environment.
AI, Geopolitics, and Digital Sovereignty
At the same time, the macroeconomic environment tested governments and businesses, as tariffs and geopolitical tensions prompted a reassessment of supply chains and business models. This led many companies to redesign their production capacities, with a significant impact on the IT ecosystem as well.
Although the AI narrative largely remained anchored to internal efficiency, 2025 marked a substantial broadening of its scope: organizations began relying on AI to manage uncertainty, dynamically update economic and financial forecasts, adapt supplier portfolios, and predict business performance in highly volatile scenarios. Against this backdrop, digital sovereignty emerged as a strategic priority, as evolving European regulations and growing awareness of jurisdictional issues prompted companies and providers to reassess their architectures, data, and processes.
Based on major global analyst forecasts, eight technological directions can be outlined that could shape business evolution in 2026.
Multi-Agent Systems: From Agentic AI to Collaborative Systems
According to Gartner, in 2026, we will see a natural evolution from individual autonomous AI agents to multi-agent systems capable of collaborating, sharing information, and contributing to even complex workflows. The trend was already evident in 2025 with early real-world applications; in 2026, it will move from a frontier topic to a more widespread practice, though not yet mainstream.
While agentic AI allowed the automation of specific tasks, multi-agent systems elevate this logic with complex architectures where each agent is trained for a specific purpose and acts as a specialized module, while simultaneously interacting with others. For example, in sales departments, multiple agents could work in sequence:
- One dedicated to dynamic lead qualification;
- One for predictive analysis of purchase propensity;
- One for automated proposal personalization;
- And a final agent preparing sales materials.
Domain-Specific AI Models: AI at the Heart of Enterprise Processes
Analysts predict that in 2026, generic language models (LLMs and SLMs) will lose prominence in favor of Domain-Specific Language Models (DSLMs), a category of models trained primarily on specialized data and designed to meet the needs of a sector, function, company, or specific process.
These are not “Small Language Models,” but models capable of interpreting context, terminology, and operational constraints that generalist models struggle to achieve efficiently. Using specialized AI models also helps reduce hallucinations, a typical side effect of LLMs.
Forecasts suggest that by 2028, over half of the GenAI models adopted by companies will be domain-specific.
Neoclouds: the Rise of GPU-First Providers
According to Forrester’s 2026 predictions, neoclouds are expected to capture €20 billion in revenue from hyperscalers due to their verticalization in Generative AI.
These cloud providers (e.g., CoreWeave) rely on physical GPU-first infrastructure and platform services (PaaS) designed to create and manage AI workloads. They are, of course, younger players compared to global giants but are already rapidly expanding, supported by significant investments.
Digital Sovereignty: an Increasingly Central Factor
In 2026, data sovereignty will remain one of the most influential forces in IT decision-making. It is not just about compliance or infrastructure preferences: the entire geopolitical landscape, combined with evolving European regulations, is pushing companies to reconsider where data resides, which actors have jurisdiction, and which architectural models ensure operational continuity in an unstable context.
This trend concerns not only those evaluating new cloud migrations but also companies undertaking selective repatriation projects to bring critical workloads back to more controlled or regional environments. The logic is workload-first: each workload finds its optimal placement based on risk, latency, compliance, performance, and data sensitivity.
AI Sovereignty: End-to-End Control Over Data, Models, and AI Infrastructure
According to IBM, in 2026, AI sovereignty will become a critical requirement for adopting artificial intelligence systems.
As with data sovereignty, this refers to an organization’s ability to maintain full control over its AI systems, the data that powers them, and the infrastructure, far from trivial in a context characterized by global technology dependencies, computational resources concentrated in few regions, and an increasingly rigid regulatory framework.
Quantum Advantage: the Era of Collaborative Ecosystems
IBM also predicts that in 2026, so-called Quantum Advantage will be reached, meaning quantum computers will be able to solve problems with measurable benefits over classical models in terms of accuracy, speed, and cost.
Despite recent accelerations, analysts emphasize that quantum computing is not a technology a company can manage independently, which is why we will see increasingly structured partnerships among businesses, cloud providers, and research centers.
Attention will also continue to grow in post-quantum security: spending on quantum security is expected to exceed 5% of the total IT security budget.
Hyperautomation Holds its Ground
Despite global attention on generative AI, systematically automating as many business processes as possible (hyperautomation) will remain central to IT strategies. The reason is clear, and there are no significant changes compared to previous years: companies know that no single technology, however powerful, can alone match the combined potential of integrations, workflow orchestration, RPA, and agentic AI, which will attract the majority of research investment.
Physical AI: the Intelligence Enters the Physical World
According to Gartner analysts, in 2026 the relationship between artificial intelligence and advanced robotics will become increasingly close. After years of perceiving AI as purely software technology, we will enter a phase in which cognitive, sensory, and decision-making capabilities are integrated directly into machines, devices, and operational infrastructure.
The push toward convergence arises from the need to make physical processes more resilient, especially in contexts characterized by demand volatility, skill shortages, and strict safety constraints.
AI, but not only AI
In 2026, artificial intelligence will continue to make headlines, likely with a different focus than in the past. The narrative will not replace the excitement about AI’s innovative potential but will highlight issues such as governance, security, sovereignty, reliability, and alignment with business processes.
At the same time, 2026 will likely not be an AI monologue. Other technological trends will begin to occupy their deserved space: new cloud paradigms driven by sovereignty needs; advances in quantum computing and its competitive advantages; the growth of Physical AI, redefining human-machine interaction; and finally, an increasingly stringent cybersecurity and regulatory landscape that will compel companies to integrate resilience and compliance into all technology decisions.
