Cohere’s European push highlights the rise of AI’s middle powers beyond the US and China

Cohere and Aleph Alpha announced a strategic transatlantic alliance, with Schwarz Group committing $600 million in future financing and set to lead Cohere’s upcoming Series E round. The deal underscores accelerating sovereign AI efforts outside the U.S. and China, as middle powers seek local AI infrastructure, data control, and alternative model ecosystems. Mistral’s prior $2 billion round at a $14 billion valuation and ASML backing further highlight growing capital formation in the non-U.S./China AI stack.

Analysis

This is less an AI adoption story than a fragmentation story: sovereign-capable buyers are increasingly willing to pay a premium for jurisdictional control, even if model quality lags U.S. frontier labs. That shifts the competitive battleground from pure benchmark performance to procurement, compliance, and deployment flexibility, which advantages vendors that can monetize on-premise, air-gapped, and data-residency use cases. The second-order effect is that the addressable market for non-U.S. leaders expands fastest in regulated industries and public-sector workloads, where switching costs are sticky and vendor concentration risk is now a board-level concern.

The biggest winner in public markets is not the startups themselves but the infrastructure layer that enables “sovereign AI” to function without relying on American cloud hyperscalers. Semiconductor lithography and advanced manufacturing tooling remain bottlenecks: each new regional AI stack still requires more GPUs, memory, networking, power, and fab equipment, so compute nationalism likely increases capex intensity rather than reducing it. That creates a multi-year tailwind for the picks-and-shovels ecosystem, while model-layer economics remain brutally competitive and prone to dilution as governments subsidize too many national champions.

The risk to the theme is that sovereign AI becomes a procurement slogan before it becomes a durable platform strategy. If enterprise buyers discover that open-weight systems are sufficient for internal use cases, the moat shifts away from model quality toward integration, services, and distribution — areas where incumbents and local consultancies can compress margins quickly. A reversal catalyst would be a renewed cross-border investment cycle or a U.S. détente on AI export controls, which would reduce the urgency premium embedded in non-U.S. vendors over the next 6-18 months.

The contrarian view is that the market is underestimating how much this trend benefits select U.S. and allied suppliers indirectly. Even when countries diversify away from U.S. model providers, they still need NVIDIA-class compute stacks, ASML-enabled fabrication, and Western enterprise software to operationalize those systems. In other words, the sovereign AI push may diversify the branding layer while further entrenching the same upstream chokepoints.

Key Decisions for Investors

• Long ASML on a 6-12 month view: sovereign AI increases global fab and tooling demand irrespective of which model vendor wins; use pullbacks to build a position, with a favorable risk/reward if AI capex remains even modestly above consensus.
• Initiate a relative-value long ASML / short BABA pair trade: the article’s negative read-through is stronger for China-facing platform names than for semiconductor enablers; the pair isolates geopolitics-driven divergence with limited macro beta.
• Buy 3-6 month call spreads on major U.S. compute suppliers / AI infrastructure names into any policy-driven dip: the second-order beneficiary is sustained capex, and the upside is capped less by model competition than by supply constraints.
• Avoid chasing late-stage sovereign-AI private-markets exposure at headline valuations: model-layer winners are likely to see subsidies and strategic capital, but margin durability is low and the exit path is uncertain over the next 12-24 months.
• Monitor for a reversal catalyst in the next 1-2 quarters: any U.S.-allied policy softening or cross-border cloud partnership would reduce the urgency premium; trim sovereign-AI-themed longs if procurement rhetoric outpaces signed enterprise deployments.