Europe Has Lost Race for Frontier AI Models, Says Rhine Group's Garicano

Rhine Group Executive Director Luis Garicano said Europe has lost the race to develop frontier AI models, highlighting a competitiveness gap with leading global AI powers. He argued Europe can still capture value through AI implementation, as the Mario Draghi-backed Rhine Group convenes economic, industry and finance leaders to address the region's competitiveness challenges ahead of Ursula von der Leyen's State of the Union address.

Analysis

The investable implication is not a broad European AI deficit trade; it is a shift in where Europe can monetize AI. Frontier-model economics remain concentrated in US hyperscalers and semiconductor supply chains, while European value creation is more likely to accrue to implementation vendors with distribution into regulated enterprises: SAP, Capgemini, Accenture and industrial software/automation franchises such as Siemens and Schneider Electric. Their upside depends less on model ownership than on converting fragmented legacy IT, data-governance and workflow complexity into recurring services and software revenue.

Near term (days to 1-3 months), political rhetoric is unlikely to move estimates without procurement, energy-policy or capital-market follow-through. The critical catalyst is whether EU competitiveness initiatives translate into funded sovereign/defense cloud, grid, data-center and industrial digitization programs; these would favor European systems integrators and electrification suppliers, while raising demand for US compute hardware indirectly. The main risk is that implementation budgets remain constrained by weak European corporate profitability and elevated power costs, leaving AI pilots unable to clear ROI hurdles.

The contrarian point is that Europe’s apparent weakness in frontier models can be an advantage for adopters: open-source and API-based models reduce the need for local model-development capex, allowing enterprises to capture labor-productivity gains faster than they can build proprietary models. That outcome is bearish for the notion that every European AI beneficiary must own GPU infrastructure, but bullish for vertical software and consulting firms that can demonstrate booked AI-related backlog and margin-accretive delivery utilization over the next 6-18 months.

Key Decisions for Investors

• Prefer a 6-12 month implementation basket: long SAP / Schneider Electric / Siemens against a short EURO STOXX 50 hedge. Target beneficiaries with enterprise distribution and pricing power; reassess if FY2026 guidance shows no AI-related backlog or margin contribution.
• Use Capgemini as a watch-item rather than an outright long until booking data improve: upside is meaningful if European enterprise AI projects move from pilots to managed-service contracts, but weak discretionary IT spending can overwhelm the theme. Trigger on sequential improvement in bookings and utilization.
• Maintain exposure to US AI infrastructure through a quality basket or SMH rather than betting on a standalone European frontier-model champion. European implementation demand is incremental to global compute demand, but the trade is vulnerable if hyperscaler capex guidance rolls over.
• Monitor EU budget/procurement announcements over the next 1-3 months for defense cloud, grid modernization and sovereign-compute commitments. Funded programs would strengthen the long Siemens/Schneider thesis; unfunded policy language is not a catalyst and should not justify adding risk.