(Moderator: André Villarroel — Chief AI Officer, Innoverse.ai)

Opening: from “better-faster-cheaper” to an Intelligence Economy

André sets a playful, multilingual tone (Spanish ↔ German ↔ English) and reframes “infrastructure” as the substrate of society, not just hardware. Drawing on World Bank fieldwork and his “Iron-Man exoskeleton for micro-economic development,” he spotlights a generational shift: youth already “vibe-coding” their way to augmented creativity—and doing it outside expensive frontier stacks. The question for Switzerland and Europe: how do we build institutions and infrastructure fit for this new intelligence economy?

In order of appearance

1) Intelligence Economy

Dr. Andrei Villarroel — Chief AI Officer, Innoverse.ai

• Thesis: The internet’s distribution era is yielding to AI’s production era—tools that make tools, code that writes code. Institutions built for industrial workflows won’t survive contact with this reality; they must be redesigned, not merely digitized.

• Equity & access: In Latin America workshops, no one touched “frontier AI” because $200/mo is untenable. Yet everyone resonated with “exoskeleton” augmentation—micro-entrepreneurship powered by small, local, affordable AI.

• Education as critical infrastructure: Move from memorization to orchestration—remix, co-create, and ship. The talent is ready; the system isn’t.

2) Sustainable Infrastructure

David Galbraith — Partner, F2N

• Network effects have moved:

  • Internet: value aggregated in user networks.

  • AI: value concentrates in production networks (software making software; robots building robots) → capital flows upstream into compute, energy, robotics, and data-centers.

• Market structure (ex-China): four LLM constellations—OpenAI/Microsoft, Anthropic/Amazon, Google, and xAI/Oracle—“all powered by NVIDIA.”

• Strategic doctrine: Sovereignty shouldn’t be defensive; win with better models, regulate after, then redistribute.

• Open models as industrial strategy: “Browser playbook” redux—open commoditizes the layer you don’t monetize, shifting profit to where you do (e.g., manufacturing).

• Europe’s wedge: Life sciences and robotics. Germany’s social pact requires productivity gains from AI-robotics (wealth creation even if net jobs don’t grow).

• Energy realism: If emissions ≈ 0, the constraint is cost not consumption. With >90% low-carbon grids, Europe should host the world’s clean compute (data-centers as a climate advantage, not a liability).

• Safety inversion: “Nice AIs” that manipulate are riskier than ones that merely refuse—sycophancy as the real social hazard.

3) Resilience

Philippe Van Caenegem — Co-Founder & Partner, The Evident Project

• Hidden architectures of civilization: Roman roads → British sea lanes → American grids → today’s cognitive grid. It will reshape everything—and risks extreme concentration of power if unexamined.

• Counter the fear script: Job-loss panic is often a tactic to freeze challengers. The real work is redistribution of the new wealth and embedding AI in the physical world (roads, labs, factories, cities).

• European playbook: Europe excels at public-private mega-projects and industrial clusters (think CERN). That is exactly what the cognitive grid demands.

• Narrative & education: We need a Swiss/European story anchored in wisdom (not just knowledge). Education must become multidisciplinary orchestration, not siloed memorization.

• Provocation: “Robots will finance our retirement.” The social contract must say how.

4) AI-Native Companies

Sal Matteis — Author, AI [r]ecursive; former Yahoo exec, investor

• Humanity OS: Technology is a tool; values are the operating system. Judge AI by the human outcomes it enables.

• Leadership myopia: Politicians chase hype (e.g., splashy gov’t deals with frontier providers) and CEOs act as gatekeepers optimizing quarters—not civilization.

• VC naïveté (of vision, not returns): Capital allocates well to firms but poorly to futures. We need narrative leadership from Switzerland/Europe—design-led, ethical, long-horizon.

• Institutional obsolescence: Education, NGOs, governments are industrial-era artifacts; avoiding hard tradeoffs breeds populism. Only technocratic recalibration—with legitimacy—can realign timelines.

Cross-panel debate: small vs. “Stargate” AI, and who gets the cognition?

• Demand-side equity: David argues the flywheel turns when everyone can afford the product (Warhol’s “same Coke” effect), pushing capital to redistribute and pushing energy costs down (cheap, clean power = democratized cognition).

• Governance warning: Philippe counters that cognition isn’t a uniform Coke—centralized models can ration quality, amplifying inequality of thinking power. Hence the need for diversity of models, swappability, and transparent allocation.

• Market transition: Sal notes cost/token is falling, but market power and bundling still set end-user prices during the transition; sovereignty requires policy + procurement aligned with long-term outcomes.

Institutions to remake (not break)

• Education: From subject silos to studio labs where students orchestrate models, data, tools, and people. Measure adaptation velocity, not rote recall.

• Energy + Compute: Treat clean compute as strategic industry: co-site renewables, heat-recovery, grid services, and sovereign DC operations.

• Robotics: Pair foundation models with embodied systems; productivity gains underpin Europe’s social pact.

• Public procurement: Buy open, auditable systems; require swappability (no single-vendor captivity) and open data by default.

The curator’s close: from Citizen-first to Planetary-centric

Picking up Otto Scharmer’s ladder (Society 1→4), Raphael Briner (curator) floats Society 5: Planetary-centric—citizen and planet co-prioritized.

What Switzerland & Europe should do next (concrete moves)

1. Clean-Compute Acceleration: Fast-track permits and PPAs for low-emission data-centers; integrate heat-reuse; publish €/token & gCO₂/token dashboards.

2. Sovereign Model Garden: Fund auditable, open LLMs (and small specialized models) with hard swappability requirements across vendors and clouds.

3. Cognitive Grid Pilots (City Labs): Run urban testbeds (health, mobility, permitting) with human-in-the-loop decisions, open data pipelines, and public post-mortems.

4. AI-Robotics Industrial Clusters: Anchor in life sciences and manufacturing; co-invest in test lines, simulation twins, safety benches.

5. Education OS Update: Replace capstone theses with capstone deployments; mandate cross-disciplinary studios (policy × design × code × operations).

6. Sovereignty by Procurement: “Open-by-default, swap-by-design” clauses; lifecycle TCO tied to energy intensity; forbid irreversible lock-in.

7. Redistribution Mechanisms: Tie public co-investment to dividends for citizens (skills credits, compute vouchers) so productivity → shared prosperity.

Memorable lines & motifs

• “Software that writes software; robots that build robots.”

• “If emissions are zero, the constraint is cost—not consumption.”

• “The cognitive grid will be our new operating system.”

• “VCs aren’t naïve about money; they’re naïve about the future.”

• “Robots will finance our retirement—if we design the redistribution.”

• “Sycophancy is a safety risk.”

One-page takeaway

Industrial → Intelligence Economy. Network effects moved from users to production. Europe’s edge is clean energy, open/auditable AI, life-sciences, and robotics—if matched by institutional redesign. Build the cognitive grid as a public-private mega-project with swappability and planetary-centric governance. Teach orchestration. Procure for sovereignty. Pilot in cities. Redistribute the gains.