If you read only one Davos debrief this year, read this one.

At World Economic Forum at Davos this year, AI was everywhere.

But AI wasn’t the main character.

Capital was.

(AI Generated)

Across sessions, whether the topic was AGI, geopolitics, markets, or humanity, the same chain kept reappearing, sometimes explicitly, often indirectly:

• AI is changing everything.

• AI needs compute.

• Compute needs energy.

• Energy needs infrastructure.

• Infrastructure needs capital.

• And capital, unlike narratives, has constraints.

What struck me most was not the optimism about AI’s capabilities, but how often progress depended on things that are slow, physical, and expensive: grids, chips, land, copper, skills, trust, governance. The future is not blocked by intelligence. It is blocked by allocation.

Several patterns repeated themselves:

AI is no longer framed as a productivity experiment, but as a macro-force: strong enough for institutions like the IMF to revise outlooks, yet still unevenly diffused into the real economy. Financial markets may be pricing growth, but adoption remains the real bottleneck. Without integration into enterprises, AI becomes a valuation story rather than a growth engine.

The race is quietly shifting from "who builds" the best models to "who can finance", power, and scale ecosystems. Compute is becoming the gatekeeper of access to intelligence. Energy advantage is turning into strategic advantage. Capital intensity is rising, not falling.

There was also an unspoken tension: going fast without guardrails versus moving too slowly to matter. “Human in the loop” has already evolved into “human in the lead”, a recognition that governance, labor adaptation, and trust are now economic variables, not ethical side notes.

One observation form Gilles Moëc, AXA Group Chief Economist, stayed with me: Europe may not lead the AI race in models or infrastructure, but it could still lead the race of adoption. That, economically, may matter more than we admit. This is a powerful message.

Conclusion

The Davos takeaway for me is simple but uncomfortable: AI will not create an age of abundance by default. It will amplify whatever capital structures, energy advantages, and coordination capacities we already have.

In closing, I’d like to echo the words of Kristalina Georgieva, Managing Director of the IMF: “Do not fall into complacency — growth is not strong enough.”

The question is no longer what AI can do; it is who can afford to make it real — and at what scale.

______________________________________________________________

Sessions that shaped this interpretation:

• Scaling AI: Now Comes the Hard Part

• The Day After AGI

• Chief Economists Briefing: What to Expect in 2026

• An Honest Conversation on AI and Humanity

• AI Power Play: No Referees

• Are Markets Mispricing the Future?

• Decade Déjà Vu: Are the 2020s the New 1920s? (popularly called "doom&gloom panel")

• Building AI for the Long Term (*my absolute favorite!*)

• Global Economic Outlook (Closing Panel)