Web Summit 2025: AI Takes Centre Stage, Robotics Becomes Physical, and Quantum Leaves the Lab

Web Summit 2025 brought 71,386 participants, 2,725 startups, 1,857 investors and 869 speakers from 157 countries to Lisbon. Yet the most important message was not the scale of the event. It was the convergence of technologies and strategic constraints that were once discussed separately.

AI agents, edge intelligence, robotics, quantum engineering, accelerated computing, energy capacity and digital sovereignty now form one system. Organizations can no longer develop a technology strategy without considering the infrastructure and geopolitical foundations beneath it.

Intelligence becomes infrastructure

AI dominated every track because it is no longer a vertical technology category. It increasingly shapes software, industrial operations, healthcare, mobility, public services and scientific research.

This changes where competitive advantage sits. Models are becoming more accessible, while the scarce capabilities are trusted data, compute, energy, integration, domain expertise and the ability to move from experimentation to reliable operations.

For executive teams, AI portfolio decisions should therefore be connected to infrastructure strategy, operating-model redesign and governance. A collection of disconnected pilots cannot produce a durable capability.

The agentic era meets the edge

Discussions across the AI tracks converged on the rise of agents: systems that do more than answer questions and can coordinate workflows, use tools and own parts of a decision loop.

The opportunity is significant, but so are the constraints. Agentic systems need explicit authority, observability, security boundaries and a method for human intervention. Their cost and performance also depend on where inference runs.

This reinforced the shift toward edge intelligence. Running models on devices, vehicles or industrial systems can improve response time, privacy, resilience and cost. Edge AI is becoming necessary wherever connectivity is variable, data is sensitive or physical action demands immediate decisions.

Robotics becomes physical AI

Amazon Robotics described a fleet exceeding one million robots and the use of AI to improve navigation and logistics. The significance is not simply fleet size. It is the progression from scripted automation toward systems that perceive, adapt and collaborate.

Foundation models for robotic navigation, improved tactile sensing and multi-arm coordination point toward greater dexterity. Multimodal and conversational interfaces could also make machines easier for workers to direct and understand.

But physical AI raises the bar for evidence. A software error can become a movement, collision or operational interruption. Deployment requires safety architecture, robust exception handling, worker trust, maintenance and an economic model tied to completed tasks.

Quantum moves from hype to engineering

Quantum computing had a smaller physical footprint at the event, but the discussion had matured. The focus moved from broad claims toward engineering: how to scale hardware, connect modules, manage errors and build software that can exploit different architectures.

Trapped-ion systems and chip-based control illustrate this industrialization path. Early work in drug discovery, materials and aerodynamic optimization suggests where hybrid classical and quantum workflows may create value first.

Broad deployment remains uncertain and likely distant, but organizations in simulation-intensive sectors should begin building quantum readiness now. That means identifying problems, creating partnerships, training talent and designing software that does not depend on one hardware path.

Sovereign compute and energy become strategic

The largest announcement around the event came from Microsoft: more than $10 billion planned for AI infrastructure in Portugal, including 12,600 next-generation NVIDIA GPUs in Sines with Nscale and Start Campus.

The announcement matters beyond Portugal. It signals Europe's need for industrial-scale AI capacity and a more distributed global compute landscape. Sovereignty increasingly depends on practical access to processors, data centers, networks, energy and the skills required to operate them.

AI infrastructure also exposes an energy constraint. Data-center power, cooling, grid availability and water use influence both cost and speed of deployment. Sustainable AI cannot be treated as a reporting layer added after architecture. Resource efficiency must become a design criterion.

This convergence reshapes geopolitics. Regions will compete for compute capacity, energy, talent and resilient supply chains in much the same way industrial powers competed for strategic resources in earlier eras.

Executive implications

Web Summit 2025 made one thing clear: innovation is becoming systemic. Software, hardware, energy, science, policy and industrial capability increasingly determine one another.

The organizations that understand this architecture of intelligence will make better technology choices and build more resilient competitive advantage.

Sources and further reading

1. Original LinkedIn article.
2. Web Summit 2025 by the numbers.
3. Microsoft: AI infrastructure investment in Portugal.