The European data-centre pipeline is expanding faster than the infrastructure required to energise it—a tension Altherum examined in the first chapter of this series. For private-market investors, the next question is narrower: which specific projects control the resources, contracts and execution capabilities required to become operational and create value?
The difference between a deliverable opportunity and an announcement can be reduced to five practical tests.
1. Is the power secured—or merely requested?
A site near abundant generation is not necessarily power-ready. The analysis begins with the actual grid position: signed connection agreements, available capacity, energisation date, redundancy, expected curtailment and the party responsible for network upgrades. A place in an interconnection queue is not equivalent to deliverable power.
The quality of the power solution matters as much as its quantity. Price visibility, duration, carbon intensity and resilience can influence customer demand, operating margins and the long-term relevance of the asset.
2. Does the location fit the workload?
AI training, inference, cloud services and latency-sensitive applications do not share the same location requirements. A lower-cost site in a secondary market may suit one workload and be unsuitable for another. Fibre routes, latency, access to engineering talent, permitting, climate, water availability and proximity to end users should be assessed as one system.
The right question is not whether the location is fashionable. It is whether the workload and the site genuinely fit.
3. Is demand contracted—or simply forecast?
Sector growth does not guarantee revenue for every facility. Investors need to understand who will occupy the capacity, when commitments begin, how prices can change and what happens if a customer delays or reduces deployment.
Pre-leasing can strengthen visibility, but it also creates concentration risk. A credible assessment therefore considers tenant quality, contract duration, renewal and termination terms, credit support, expansion rights and the cost of adapting the asset to another user.
4. Can the project actually be delivered?
Land and power do not build a data centre. Delivery depends on permits, design, construction management, access to transformers and other long-lead equipment, cooling architecture, cybersecurity, commissioning and an operator capable of running a high-density environment.
The schedule should be tested against dependencies rather than accepted as a single target date. Milestones, contingencies and accountability reveal whether the plan can absorb delays without destroying the investment case.
5. Can the capital structure survive delay?
AI infrastructure is capital intensive and timing risk is real. The financing plan should therefore be examined under a slower build, a later energisation date, higher construction costs and a more gradual occupancy ramp.
For an equity club deal, the analysis includes the entry valuation, equity cushion, staged capital calls, governance rights, dilution protections, decision thresholds and exit routes. Project-finance debt—whether structured as direct lending, club debt or infrastructure bonds—requires a separate assessment of the borrower or issuer, security package, covenants, repayment waterfall and pricing. Equity and debt are not interchangeable merely because they finance the same theme.
What the club-deal structure changes
A club deal does not transform enthusiasm for AI into investment quality. Its value is that it can focus capital on one identifiable project or operating company, with defined governance, milestones and economics. Investors can evaluate the asset itself rather than buying exposure to a broad narrative.
That focus is particularly relevant when the opportunity is too large or operationally complex for an investor to pursue alone, but sufficiently specific to be analysed transaction by transaction. The discipline remains the same: power, location, demand, delivery and capital structure must support one another.
The best AI-infrastructure opportunities will not necessarily be the ones with the loudest technology story. They will be the ones able to turn electricity, land, contracts and execution into durable operating capacity.
Internal link: AI has a physical address: how infrastructure can become a club-deal opportunity