Sovereign Stacks: Building Trusted AI on National Terms | RAISE Summit 2026

Panel Definition of Sovereignty: The consensus defines sovereign AI not as autarchy or isolation, but as pragmatic organizational control, choice, and autonomy over data, workflows, and infrastructure.

• Gleb Budman (Backblaze) notes that sovereignty is often confused with mere "residency" or "security," emphasizing that the goal is the ability to move data, select access points, and avoid egress fees that tax sovereignty.
• Simon Musket (Digital Realty) argues that sovereign decisions regarding data location and access ripple through the entire infrastructure stack, affecting performance, cost, and network architecture.
• Thomas Di Giacomo (SUSE) identifies control and vendor exit options as the core of sovereignty, distinguishing it from simple regulatory compliance.

Stack Layer Relevance and Trade-offs: The panel agreed that while all layers of the AI stack matter, sovereignty requirements vary by industry use case and risk profile.

• Mohsen Moazami (DDN) observed that defense and telecom sectors require sovereignty across the entire stack, whereas lower-stakes sectors like hospitality may prioritize cost over full sovereignty.
• Corey Hochveld (HyperStack) highlighted that AI introduces new complexities where traditional "crown jewels" (data) pass through models, breaking traditional security and infrastructure models.
• The panel noted that forcing complete geographic or vendor residency can hinder innovation if it restricts access to superior global hardware or software.

Geopolitical Impact on Innovation: Constraints and export controls are viewed as catalysts for technological advancement rather than purely negative barriers.

• Mohsen Moazami cited the necessity for China and the Middle East to develop domestic models and hardware due to U.S. restrictions, noting that Saudi Arabia is now exporting AI tokens.
• Corey Hochveld argued that U.S. foreign policy and restrictions on Chinese access to NVIDIA GPUs have artificially created competition, forcing domestic innovation in hardware and software.
• Simon Musket emphasized that while the U.S. and China lead the race, Europe must focus on leveraging its own engineering talent and startups to avoid falling further behind.

Scale, Economics, and Infrastructure: The scale of AI demand is shifting from commodity procurement to complex financial and logistical decision-making.

• Mohsen Moazami contrasted European "neoclouds" planning for tens of thousands of GPUs with a U.S. entity securing a single $100 billion purchase order, highlighting a massive gap in deployment scale.
• Simon Musket explained that decisions regarding GPU placement now depend on tokenomics, liquid cooling requirements, latency, and network efficiency, representing hundreds of millions to billions in capital decisions.
• Mohsen Moazami stated that average AI infrastructure sits idle 40–60% of the time, making infrastructure productivity and utilization rates the primary metric for profitability, not just raw capacity.

Europe's Strategic Path Forward: The panel suggested Europe should avoid mimicking U.S. hyperscalers and instead focus on open ecosystems and industrial application.

• Thomas Di Giacomo warned that attempting to rebuild domestic equivalents of AWS in Europe is a waste of time due to the speed of regulation and obsolescence.
• Di Giacomo advocated for an open ecosystem where European vendors partner globally (with NVIDIA, Chinese providers, and French clouds) to deliver high-performance ROI to end businesses.
• Mohsen Moazami urged European leaders to stop focusing on "regulating" and start capturing the massive commercial demand for compute that is currently going elsewhere.

Future Outlook and Key Myths: The concept of sovereignty is expected to remain relevant through 2030–2035, though the terminology may evolve from "sovereignty" to "resilience" or "interdependence."

• Myth 1: Sovereignty degrades capability. (Dismissed by Corey Hochveld, who asserts sovereign solutions can meet enterprise needs without relying solely on frontier U.S. models.)
• Myth 2: Sovereignty equals residency. (Dismissed by Gleb Budman and Simon Musket, who emphasize that data utility depends on the ability to move and process data, not just its physical location.)
• Myth 3: Regulations alone drive sovereignty adoption. (Dismissed by Thomas Di Giacomo, who argues that business needs and technology economics will ultimately dictate adoption, not just legal mandates.)