Davos 2026: AI is reshaping global power and infrastructure.

At Davos 2026, leaders framed artificial intelligence as a large, capital‑intensive infrastructure driving economic and geopolitical shifts; discussions focused on governance, control and limits in compute, energy and data center capacity.

AI as infrastructure is the central theme coming out of Davos 2026. Delegates described AI not as a mere tool but as a physical, capital‑intensive buildout tied to power grids, chips and data centers. Conversations moved from whether AI will change the world to how control, governance and timing will be decided. The event highlighted both confidence in AI’s economic potential and concern about uneven benefits and institutional readiness. Key takeaways: - Many at Davos framed AI as an infrastructure project linked to compute, energy, chips and data centers rather than only software. - The article reports concerns about economic imbalance, productivity divergence and who benefits from AI-driven change. - Former OpenAI board member Helen Toner told U.S. Congress that human‑level AI could arrive within one to three years and could pose existential risk, a point the piece notes was lightly covered by mainstream news. - A BCG presentation cited that 72 percent of CEOs see themselves as primary decision‑makers on AI, and many expect AI agents to deliver positive ROI by 2026. - The piece reports that some leaders are moving from integration toward selective cooperation and that alliances are increasingly following technology stacks; Europe was described as facing harder choices due to regulation and fragmentation. - The article notes infrastructure limits—training compute growth, inference cost shifts, energy supply, grid reliability, payment rails and data center capacity—are now central constraints on deployment. Summary: Davos 2026 presented AI as a force reorganizing economic and geopolitical influence around compute, energy and talent. The immediate impact is both organizational—CEOs taking primary responsibility for AI—and structural, with infrastructure capacity shaping how quickly AI moves from demonstration to reliable systems. Undetermined at this time is how governance and responsibility will be aligned with the rapid pace of investment and deployment.