From NVIDIA's AI factories to the EU AI Act's command-center approach, this panel maps the infrastructure layer that makes enterprise AI possible — or impossible. An NVIDIA public sector lead, a Microsoft Asia architect, a Visium CEO, and the architect of Europe's AI regulation debate what it really takes to move from isolated pilots to scalable, interoperable AI systems.

WHAT THIS PANEL COVERS

• NVIDIA frames AI factories as the industrial revolution's new foundational asset — importing electricity and data, exporting intelligence — with sovereignty as a core design principle
• The biggest enterprise mistake is focusing on isolated point solutions instead of building shared, modular platforms that make AI and data reusable across the organization
• The EU AI Act was designed as a governance 'command center' but expanded into legal uncertainty — even its architect admits he does not fully understand all parts of the regulation
• Platform interoperability is the critical enabler: protocols like MCP and A2A agent standards are the 'electricity and water systems' connecting AI's isolated buildings into a functioning city
• Leadership must lead by example — being the first to use AI, ask for AI-driven insights, and invest in AI literacy — or the organization will treat AI as something imposed rather than embraced

PANELISTS

• Stephan Balzer — Keynote Speaker, Host & Moderator; Managing Director, red onion (Moderator)
• Eva-Maria Hempe — Public Sector Lead, NVIDIA
• Alen Arslanagic — CEO, Visium
• Kai Zenner — Head of Office for MEP Axel Voss; Digital Policy Adviser, European Parliament
• Ujjwal Kumar — Principal Architect, Office of CTO, Microsoft Asia

unDavos is a community-driven summit running during WEF week in Davos, democratizing the conversation around global challenges.

🌐 undavos.com

Tags: AI infrastructure, AI at scale, NVIDIA, EU AI Act, sovereign AI, AI factories, interoperability, AI platforms, enterprise architecture, AI governance, European Parliament, Microsoft Asia, Davos 2026, unDavos, REWIRE