Is AI the new critical infrastructure

AI: The New Global Infrastructure – Power, Control, and Rapid Evolution at Web Summit Lisbon 2025

The Web Summit Lisbon 2025 panel debated if AI is the new critical infrastructure, questioning its role as a neutral utility or a source of power. Mr. Lars Gehrmann of Qatar Insurance Group believes AI will become a utility, but its rapid evolution means projects quickly become obsolete, demanding constant adaptation.

Mr. Tiago Henriques from Google Cloud highlighted Google’s $75 billion AI infrastructure investment, supporting a core platform for global enterprises. Mr. Waqar Aziz of Microsoft described a three-layered foundation: physical infrastructure by providers, sovereign guardrails by governments, and consumer adoption.

Microsoft is shifting to “AI superfactories”—interconnected data centers focused on sustainable growth, measuring “kilowatts per hour per dollar.” Mr. Henriques detailed Google Cloud’s GPUs and proprietary TPUs, powering models like Gemini, and offering open models like GEMA for customer flexibility.

Mr. Gehrmann noted that for Qatar Insurance Group, AI isn’t yet full infrastructure due to regulatory limits on local high-performance computing. This restricts full-scale production, posing a challenge for regulated industries reliant on global providers.

“Sovereign AI” was a key topic. Mr. Aziz defined it broadly, beyond data residency, to include choice, resilience, auditability, and legal coverage. It grants control over data, operations, and infrastructure, ensuring geopolitical stability and incorporating regional cultural values into LLMs.

Mr. Henriques confirmed sovereign AI as a priority, offering flexible public, distributed, and air-gapped cloud solutions. These provide varying control levels, allowing customers to deploy AI innovations, including Gemini, based on specific use cases and data sensitivities.

Mr. Gehrmann stressed that AI’s rapid development, with capabilities changing monthly, challenges corporates. Organizations must continuously re-engineer processes and adapt business models to leverage new agentic AI, like “digital twins,” for competitive advantage.

Regarding funding, Mr. Aziz suggested public-private partnerships for AI infrastructure due to high capital expenditure. Public sector involvement increases as AI diffuses into critical applications. Mr. Henriques emphasized Return on Investment (ROI), noting over 80% of MENA executives saw tangible AI ROI, rapidly deploying generative AI solutions.

Looking five years ahead, Mr. Gehrmann predicted AI infrastructure would be a source of geopolitical power and competition, like 20th-century oil. He highlighted AI’s profound, accelerating integration into daily life. Mr. Henriques anticipated models becoming significantly smarter, performing complex tasks efficiently across science, drug discovery, and robotics.

Mr. Aziz concluded by emphasizing inclusion and trust. He highlighted Microsoft’s commitment to community impact, ensuring data center operations do not burden local electricity grids and that job opportunities benefit local communities. Responsible AI development requires conscious inclusion and maintaining trust.