Anthropic looks to hire six-figure role for negotiating data center deals to fuel Europe AI expansion

Anthropic is moving to secure European data center capacity, hiring a London-based Transaction Principal to source and negotiate deals across FLAP-D and other European markets. The company recently committed to spending more than $100 billion on Amazon Web Services over 10 years and signed an expanded Broadcom deal for about 3.5 gigawatts of computing capacity. The news underscores accelerating AI infrastructure demand in Europe, though it is still early-stage and not yet a signed Europe deal.

Analysis

The signal here is not just incremental demand for European capacity; it is a forced repricing of scarce, power-secured land banks. The fastest beneficiaries are the vendors that can package permitting, power access, and financing into a single executable contract, which structurally favors scale players over pure colo names. Broadcom is the cleanest second-order winner because every incremental gigawatt commitment hardens the need for custom networking, interconnect, and silicon-heavy rack architecture, while Microsoft remains the broader ecosystem toll collector as hyperscalers increasingly monetize capacity before they monetize model usage.

The more interesting read is that Europe is becoming a segmentation market: cheap-power Nordics and Iberia will win the long-duration load, while FLAP-D hubs risk becoming premium latency nodes with lower utilization and weaker economics. That matters because it creates a barbell in landlord returns—developers with captive renewable or grid capacity should see higher forward multiples, while those reliant on merchant power or slower interconnect timelines could face spread compression. The likely second-order effect is a bidding war for operational sites rather than greenfield projects, which should lift residual values for existing AI-ready shells more than headline capex suggests.

Risk is mostly execution timing rather than demand; the next 1-3 months should bring more headlines, but the real cash-flow impact is 12-36 months out as leases, power purchase agreements, and chip deliveries synchronize. The key reversal catalyst is policy or power economics: if European regulators tighten around grid allocation, water use, or cross-border data rules, these deals can slip from strategic to uneconomic quickly. A second risk is that capacity gets overordered relative to model demand, setting up a 2027 digestion phase with underutilized assets and renegotiated pricing.

The consensus may still be underestimating how much of the value accrues to the infrastructure layer rather than to the AI model providers. If Europe’s compute build-out becomes capital-constrained, the best risk/reward is not chasing the obvious AI software names but owning the bottleneck providers with contract visibility and shorting the losers exposed to power scarcity and financing stress.

Key Decisions for Investors

• Long AVGO vs. a basket of unprofitable AI software names for 6-12 months: the thesis is that every new capacity deal increases content intensity per watt, with better downside protection than pure model-exposure names.
• Add MSFT on pullbacks over the next 1-3 months: use any weakness tied to capex skepticism to own the dominant toll collector on European compute distribution, with lower execution risk than direct capacity developers.
• Long NBIS only as a tactical trade, not a core position: best entry is after confirmation of contracted power and customer pre-commitments; otherwise the stock is vulnerable to project-timing slippage and financing dilution.
• Relative-value pair: long Europe power/infrastructure beneficiaries, short names dependent on merchant-grid access; hedge with a short in weaker European colo operators if available, since the market is likely overpaying for pipeline without secured electrons.
• Buy medium-dated upside in AVGO or MSFT rather than outright leverage: 3-6 month calls can capture the re-rating from additional AI infrastructure disclosures while limiting downside if European permitting slows.