A perfectly engineered data centre with no tenant is worth less than a mediocre one with a 15-year hyperscaler lease. It is how the debt capital markets price these assets, how institutional investors underwrite them, and how leading operators think about deployment risk. Yet much of the capital now flowing into the sector is being allocated as though the building were the asset. It is not. The offtake agreement is the asset. Everything else is a cost centre until it is contracted.
McKinsey estimates that data centres will require approximately $6.7 trillion in cumulative capital expenditure worldwide by 2030, according to its April 2025 analysis. The Middle East data centre market alone is expected to grow from $3.05 billion in 2025 to $7.19 billion by 2031, a compound annual growth rate exceeding 15 per cent, according to Research and Markets. Capital is abundant. The question is whether it is being deployed against the right risk.
At Claymont Equivator Infrastructure, every engagement is structured around a single principle: demand precedes capital.
Investment decks for data centre projects tend to lead with technical specifications: Tier IV redundancy, liquid cooling, high-density racks at 100 kW or more per cabinet. These matter. They are not, however, what generates revenue.
Revenue is generated by the offtake agreement. A long-term lease with an investment-grade hyperscale tenant transforms a data centre from real estate into something closer to a financial instrument, as Bryan Cave Leighton Paisner noted in a 2025 analysis. The income stream becomes predictable, bondlike, and bankable.
Without such a lease, even the most advanced facility carries the full weight of speculative demand risk. In a sector with multi-year construction timelines and capital intensity measured in the hundreds of millions, that risk is a reliable path to impaired returns.
The debt markets have understood this for some time. In a June 2025 roundtable published by Norton Rose Fulbright, one participant described hyperscaler lease-backed cash flows as close to bulletproof. ABS transactions backed by such leases have traded at approximately 100 to 150 basis points above comparable hyperscaler corporate bonds in recent conditions, a spread that reflects illiquidity, not credit risk.
Skadden observed in September 2025 that hyperscale construction financings are underwritten on a fundamentally different basis from traditional commercial real estate. The key factors are the credit worthiness of the tenant, the quality of anticipated lease cash flows, and the risk profile of the lease terms. The physical property is secondary.
The distinction sharpens at refinancing. Once a facility is operational, construction debt is replaced by permanent capital priced on tenant credit, not asset value. Debt service coverage ratio thresholds differ materially between contracted and uncontracted facilities; lenders apply tighter covenants and demand higher coverage from projects without a committed offtake. The consequence is mechanical: a contracted facility accesses cheaper, longer-dated capital. An uncontracted one pays more and borrows less.
The opposite of the contracted model is speculative construction. The evidence from mature markets suggests it carries significant risk for institutional capital.
As of late 2025, approximately €5.8 billion worth of data centre projects in Ireland were effectively stranded, with land and permits secured but no grid capacity to connect to, according to Enlit World. Similar bottlenecks are emerging in London, Amsterdam, and Frankfurt. In each case, physical assets exist. Contracted demand does not.
In the United States, the pattern is more pronounced. AIxEnergy warned in December 2025 that business-as-usual planning, treating speculative demand projections as certainties, could strand tens of billions in assets. The Rocky Mountain Institute noted that utilities have a track record of overforecasting demand, spending billions on power infrastructure for loads that never materialised. In data centre development, where a single campus may require hundreds of megawatts of dedicated power, the cost of miscalculation is proportionally severe.
The contrast in outcomes is already visible. A hyperscale campus backed by a 15-year investment-grade lease typically secures construction financing at favourable terms, proceeds through phased build-out against confirmed demand, and delivers stable returns from energisation. A comparable facility built on spec, in the same market, may struggle to attract debt, face cost overruns during an extended leasing period, and ultimately require recapitalisation.
Mawer Investment Management, analysing credit risk in the AI data centre supercycle, identified several red flags in current deal flow: projects built on speculative demand or verbal tenant interest, misalignment between lease durations and debt maturities, and facilities without fully secured power interconnections. Each traces to the same root cause. Capital was deployed before a creditworthy counterparty was secured.
The Gulf Cooperation Council region concentrates several forces that make counterparty discipline particularly consequential. GCC sovereign wealth funds collectively manage nearly $6 trillion, more than 40 per cent of the global total, [[according to Deloitte]](https://www.deloitte.com/global/en/Industries/investment-management/perspectives/gulf-cooperation-council-sovereign-wealth-funds.html). Gulf funds deployed a record $119 billion in 2025, with AI and digitalisation accounting for a significant share, [[according to AGBI]](https://www.agbi.com/analysis/finance/2026/01/gulf-wealth-funds-racked-up-119bn-of-spending-in-2025/). Saudi Arabia’s data centre market alone is projected to grow from $2.11 billion in 2025 to $4.35 billion by 2031, with installed IT load capacity expected to reach 1.03 GW by 2030, [[according to Mordor Intelligence]](https://www.mordorintelligence.com/industry-reports/saudi-arabia-data-center-market).
These strategies extend across the Gulf, not only the Kingdom. The UAE’s Stargate project, a 1 GW AI compute cluster within a broader campus in Abu Dhabi developed through a partnership involving G42, OpenAI, Oracle, NVIDIA, and SoftBank, is one of the largest single commitments to AI infrastructure globally. Qatar, Oman, and Kuwait are each advancing national digital strategies that include significant data centre capacity expansion.
The scale of capital flowing into the sector raises the stakes for every deployment decision. The discipline applied to counterparty selection must match the ambition of the investment programme.
What distinguishes the GCC from other emerging data centre markets is not grid constraints, permitting complexity, or desert cooling requirements. These are common to many developing regions. The distinguishing factor is the centrality of sovereign capital and the strategic motivation behind the build-out. In Saudi Arabia, Vision 2030 and the Kingdom’s broader Transcendence programme position the state itself as the architect of deployment. HUMAIN, a PIF-owned company delivering AI infrastructure with a reported commitment of up to $100 billion, is the principal vehicle for this ambition. Deploying capital against uncontracted capacity in this environment is structurally misaligned with the governance standards sovereign investors apply.
Saudi Arabia’s 2025 national strategy targeted 1.5 GW of data centre capacity by 2030, a figure subsequent announcements suggest the Kingdom may now exceed. Public communications from SDAIA’s national data and AI strategy indicate that sovereign and public sector workloads will anchor the national build-out, as [[Greenberg Traurig noted in January 2026]](https://www.gtlaw.com/en/insights/2026/1/saudi-arabias-data-centre-expansion-regulatory-framework-and-strategic-considerations). The infrastructure must be built against identifiable, creditworthy, strategically aligned demand. Sovereign workloads, hyperscaler deployments, and regulated enterprise requirements in healthcare, banking, and telecommunications provide it.
Access to frontier silicon adds a further constraint. US Bureau of Industry and Security export controls on advanced AI accelerators have undergone significant revision since 2023. The Trump administration rescinded the Biden era AI Diffusion Rule in May 2025 and expanded GCC access to chips previously subject to strict licence requirements, [[as reported by Data Center Knowledge]](https://www.datacenterknowledge.com/data-center-chips/ai-chip-export-controls-a-new-challenge-for-data-center-operators). The framework remains subject to ongoing policy evolution. For investors, this means counterparty agreements must account not only for power and space but for the regulatory pathway through which compute hardware will be procured.
Not all demand is equal. Four characteristics define a strong anchor counterparty.
Credit durability. Hyperscalers such as AWS, Microsoft, and Google carry investment-grade ratings. Norton Rose Fulbright noted that lenders view their long-term lease commitments as providing cash flow stability exceeding that of comparable infrastructure classes, including renewables. Leases of 15 to 20 years from such counterparties attract the deepest institutional capital at the most favourable terms.
State alignment. In the GCC, sovereign and quasi-sovereign entities, national AI platforms, and government ministries with cloud-first mandates represent demand that is structurally embedded in national economic strategy. Following the November 2025 announcement that HUMAIN would partner with AWS to deploy up to 150,000 AI accelerators in Riyadh, and with SDAIA signing strategic agreements with leading technology companies, these are not speculative signals. They are expressions of policy-backed, budget-allocated demand.
Technical alignment. The counterparty’s requirements must be deliverable by the facility. This includes power density, redundancy, cooling architecture, compliance frameworks, data residency, and localisation requirements under the Personal Data Protection Law and sectoral regulation, sovereign cloud mandates, and connectivity to international fibre routes. Misalignment between supply and demand specifications is a frequent cause of deals failing to progress.
Lease structure. Lenders scrutinise termination rights, renewal options, rent escalation mechanisms, and alignment between lease duration and debt maturity. As Clifford Chance observed in May 2025, lenders may require pre-agreed or investment grade counterparties and minimum weighted average unexpired lease term thresholds before committing capital. Weak structures, even with strong counterparties, can undermine the entire capital stack.
If the counterparty determines the value, then the counterparty should be secured before the capital is deployed. The market contains a spectrum of risk tolerance, from platforms that deploy capital against pre-leased capacity to those that build entirely on spec. The default in much of the sector is to secure land, arrange financing, begin construction, and then seek tenants. This exposes investors to the full spectrum of demand risk when capital is most concentrated and least recoverable.
The alternative inverts the sequence: identify the counterparty, confirm technical alignment, structure the offtake, then introduce capital. Investment is backed by contracted revenue rather than market assumptions. Construction follows confirmed demand, not projected demand. The facility, when delivered, has an immediate revenue path.
Platforms that originate offtake before introducing capital, engaging directly with hyperscalers, sovereign AI platforms, and enterprise buyers, structurally reduce the demand risk that undermines speculative projects. This is the approach Claymont Equivator Infrastructure follows. Capital is introduced only once demand is contractually secured, reducing demand risk for investors and structuring capacity against committed offtake.
The methodology reflects a broader institutional reality.
For sovereign wealth funds, pension funds, and infrastructure allocators, the counterparty question is not merely commercial. It is a governance requirement.
Sovereign capital demands transparency, alignment with national strategic priorities, and downside protection. A data centre backed by a 15-year lease from an investment-grade hyperscaler is structured to meet these standards. A speculative facility built on projected demand curves does not.
PwC noted that the region’s sovereign wealth funds are channelling billions into data centres as part of their diversification strategies, with both sovereign and private equity capital ensuring ample funding for state-of-the-art facilities. Capital availability is not the constraint. The question is whether projects are structured to deliver the contracted returns institutional mandates require.
In Norton Rose Fulbright’s analysis of European data centre financings, offtake contracts represent the sole source of income for a data centre project. The quality of those contracts, and the creditworthiness of the entities behind them, determines the cost of capital, the terms of debt, the recoverability of the investment, and the returns delivered to the asset owner.
The AI data centre build-out is among the largest infrastructure deployment programmes in history. Across the GCC and globally, it is attracting capital at a pace with few precedents. Scale alone does not guarantee returns. The projects positioned to deliver durable, institutional-grade outcomes are those anchored by strong counterparties with proven credit, strategic alignment, and long-duration commitments.
The asset is the contract, not the building. The contract is only as strong as the counterparty behind it.
1. McKinsey & Company, “The Cost of Compute: A $7 Trillion Race to Scale Data Centers,” April 2025. Link
2. Research and Markets, “Middle East Data Center Industry Report 2026,” February 2026. Link
3. BCLP, “Financing Data Centre Developments: Balancing Risk and Opportunity in a Capital Intensive Sector,” 2025. Link
4. Norton Rose Fulbright, “Data Center Financing Structures,” June 2025. Link
5. Skadden, “Hyperscaler Data Centers: Financing Solutions for Large Scale Projects,” September 2025. Link
6. Enlit World, “Data Centre Assets Left Stranded by Power Constraints,” late 2025. Link
7. AIxEnergy, “Managing Data Center Uncertainty Part V,” December 2025. Link
8. Rocky Mountain Institute, “Fast, Flexible Solutions for Data Centers,” July 2025. Link
9. Mawer Investment Management, “Extra Credit: Credit Investing in the AI Data Centre Supercycle,” 2025. Link
10. Deloitte, “Gulf Cooperation Council Sovereign Wealth Funds at the Forefront of a Strategic Global Expansion,” November 2025. Link
11. AGBI, “Region’s Wealth Funds Racked Up $119bn of Spending in 2025,” January 2026. Link
12. Mordor Intelligence, “Saudi Arabia Data Center Market Size and Growth to 2031,” 2026. Link
13. Greenberg Traurig, “Saudi Arabia’s Data Centre Expansion: Regulatory Framework and Strategic Considerations,” January 2026. Link
14. Norton Rose Fulbright, “Data Centre Financing: A European Perspective,” 2025. Link
15. Clifford Chance, “Innovative Datacentre Financing: Portfolio vs ABS,” May 2025. Link
16. PwC, “Unlocking the Data Centre Opportunity in the Middle East,” 2025. Link
17. Norton Rose Fulbright, “Lender Considerations on Data Centre Financings in Europe,” 2025. Link
18. Data Center Knowledge, “AI Chip Export Controls: A New Challenge for Data Centers,” December 2025. Link
This article is published for informational purposes only and does not constitute investment advice, a financial promotion, or an offer of securities. The views expressed reflect analysis of publicly available information and should not be relied upon as the basis for any investment decision.