Roughly $870 billion in data centre debt financing will be required over the next five years, according to JLL’s 2026 Global Data Center Outlook. JLL projects the global sector expanding at a compound annual growth rate of 14% through 2030, with capacity on track to nearly double to 200 GW. Private equity firms, sovereign wealth funds, and infrastructure investors are deploying capital into facilities designed to serve AI workloads for decades. They scrutinise power procurement, cooling architecture, and GPU supply chains in granular detail. They spend far less time on the layer that will determine whether those assets actually perform: governance.
The cost of this neglect is well documented. The IMF estimates that countries waste a third of their infrastructure spending, not because capital is scarce but because governance is weak. The World Bank reached a similar conclusion: infrastructure delivery problems stem overwhelmingly from governance failures, not financial constraints. In data centre development, where single projects routinely exceed $1 billion and span multiple jurisdictions, the consequences compound faster and with less visibility than in almost any other asset class.
Governance here does not mean compliance checklists or ESG disclosures. It means the architecture of decision making: who sits on the board, what they can approve without referral, how capital deployment triggers are structured, how disputes between partners are resolved, and how control rights shift as projects move through phases. These mechanics determine whether an asset performs across a 20 to 30 year lifecycle or gradually loses value through misalignment, delay, and avoidable conflict.
Governance failures in infrastructure rarely make headlines. They do not produce sudden collapses. They produce something worse: slow decay. Decisions deferred. Capital deployed at the wrong phase. Technical requirements mismatched with operator capabilities. Disputes between partners that freeze progress for months while the market moves on.
The OECD identified poor governance as a primary driver of infrastructure projects failing to meet their timeframe, budget, and service delivery objectives. This is not a public sector problem alone. A 2025 analysis by CSC found that 85% of limited partners globally reported that poor fund administration quality had negatively affected their investment decisions at least once in the previous year. In data centre investment, where joint ventures between operators, technology partners, and capital providers routinely cross jurisdictional boundaries, the governance layer is what determines whether a platform can scale or remains trapped in negotiation.
Boston’s Big Dig offers the canonical illustration. Budgeted at $2.8 billion with a 1998 completion target, it finished in 2007 at over $24 billion. The root causes were not technical. They were structural: fragmented oversight, absent long term planning, and no accountability mechanisms at the board level worth the name.
Data centre projects will not produce a Big Dig. They will produce something quieter but no less expensive: idle capacity, stranded capital, or a facility that meets every technical specification but serves the wrong market at the wrong time. The failure mode is different. The cost is comparable.
Strip away the complexity, and infrastructure governance comes down to three questions: who decides, what requires elevated approval, and what happens when the system breaks.
Board composition and appointment rights are the first. Joint venture structures in large data centre projects typically allocate board seats in proportion to equity. A 60/40 venture gives the majority partner three directors and the minority partner two. But the number of seats matters less than what each seat can do. Gibson Dunn’s joint venture governance practice has made the point directly: the critical question is which decisions require supermajority approval, which are delegated to management, and whether independent directors exist to arbitrate conflicts.
This matters acutely in projects spanning the GCC and Europe. The capital partner may be a sovereign entity with strategic objectives that extend well beyond financial returns. The operator brings technical execution. The governance framework must give both parties sufficient authority without allowing either to paralyse the other. Get the balance wrong and the venture stalls. Get it right, and it scales.
Reserved matters and veto rights are the second. Norton Rose Fulbright’s analysis identifies reserved matters as the mechanism through which minority shareholders protect their position while majority shareholders prevent smaller partners from blocking essential decisions. In standard infrastructure, reserved matters cover constitutional changes, share capital, material contracts, and key personnel appointments. Data centre projects demand more: technology refresh cycles, offtaker contract negotiation parameters, and capital expenditure triggers tied to demand milestones all require explicit treatment.
The penalty for a poorly designed reserved matters list is binary. Either too many decisions require consensus that cannot be reached, paralysing the platform. Or one party makes commitments that the other’s capital position cannot support. Both outcomes destroy value. Both are preventable.
Deadlock resolution is the third. Public company shareholders who disagree can sell their positions. Infrastructure joint venture partners cannot. The World Bank’s guidance on project company governance is clear: deadlock provisions must be agreed at formation and must include structured escalation, mediation, and arbitration. Without them, a single material disagreement can freeze an entire project indefinitely.
Governance design takes on distinct dimensions in the Gulf Cooperation Council states, where data centre investment is accelerating as part of broader national transformation agendas.
Saudi Arabia’s Communications, Space and Technology Commission (CST) introduced Data Center Services Regulations on 1 January 2024, establishing registration, classification, and quality requirements for all operators. The regulations sit within the wider context of Vision 2030’s digital infrastructure objectives and the Kingdom’s HUMAIN initiative, which is structuring a national AI compute ecosystem at sovereign scale. In April 2025, the CST released the draft Global AI Hub Law, a framework that introduces the concept of data embassies: facilities where foreign governments and private entities can host data within Saudi territory under the jurisdiction of their home country’s laws.
The draft law defines three categories of hub (Private, Extended, and Virtual), each with distinct governance and sovereignty provisions. The law, which completed its public consultation in May 2025 and remains pending as of early 2026, would make Saudi Arabia the first G20 nation with a comprehensive legal framework for sovereign data centres of this kind. The regulatory architecture is evolving at a pace, and governance structures designed today must be built to accommodate change.
The UAE is pursuing a parallel strategy. Abu Dhabi’s Mubadala and G42 have established the emirate as a distinctive node in the regional AI infrastructure landscape, while the Abu Dhabi Global Market (ADGM) and Dubai International Financial Centre (DIFC) offer distinct regulatory frameworks that international investors increasingly use to structure data centre joint ventures in the Gulf. Any governance framework operating across the GCC must account for these jurisdictional differences.
The regulatory picture is further shaped by external forces. The US Bureau of Industry and Security rescinded the Biden-era AI Diffusion Rule in May 2025, replacing a rigid tiered framework with a bilateral, country-specific approach to chip export governance. However, the underlying 2022 and 2023 restrictions on advanced chip exports remain in force, and sales to GCC jurisdictions still require US government approval. A replacement rule is pending. For governance structures in GCC data centre ventures, this means chip procurement compliance remains a live and evolving obligation that must be addressed alongside domestic regulatory requirements.
The capital dynamics are equally significant. GCC sovereign wealth funds collectively manage approximately $6 trillion, with projections placing that figure between $7 trillion and $8 trillion by 2030. These institutions have become among the most active infrastructure investors globally, with five GCC funds ranking among the top ten global dealmakers in 2024. Their engagement in data centre investment reflects not only financial return objectives but national strategic priorities: building domestic AI capability, ensuring data sovereignty, and positioning their economies within a broader diversification agenda. For large commitments, they increasingly require board representation or observer rights. They seek joint venture structures that provide direct deal access, co-investment rights, and governance input. This is a condition of capital access, not a preference.
Greenberg Traurig’s 2026 analysis of Saudi Arabia’s data centre expansion captures the practical challenge. Investments in the Kingdom routinely involve partnerships between local and international players, raising overlapping questions about ownership, licensing, governance requirements, and regulatory approvals. Each structure must satisfy local regulatory obligations while accommodating international capital and operating expertise. The governance layer is where these requirements are either reconciled or allowed to generate friction that delays deployment.
For platforms operating in this environment, governance is not a supporting function. It is a prerequisite. Sovereign and institutional partners will not commit capital to structures that lack what the market now expects: transparent decision mechanics, clear escalation paths, independent risk oversight, and capital deployment sequenced against contracted demand.
The most consequential governance failure in infrastructure is also the most common: deploying capital before the governance framework is mature enough to manage it. The pressure to move quickly, particularly when demand signals are strong, creates a persistent temptation to commit capital first and refine governance later.
CoreWeave’s March 2025 IPO illustrates the risk at the company level. The AI cloud provider raised $12.9 billion in debt commitments in roughly two years, carrying approximately $8 billion on its balance sheet by the end of 2024. It reported widening losses despite revenue growth exceeding 700%. It was forced to price its IPO well below its initial target range. Within weeks of listing, it sought an additional $2 billion in high-yield debt. The underlying technology may prove sound; the point is that the capital structure had outrun the oversight mechanisms needed to manage it, a fact underscored by the company’s own disclosure of material weaknesses in internal financial controls.
At the project level, the equivalent failure is building capacity before offtake is contracted or committing capital before the joint venture agreement has resolved basic questions about decision authority, phasing triggers, and exit provisions. Experienced infrastructure investors now insist on sequencing discipline for exactly this reason: demand secured before capital committed, governance finalised before deployment begins.
This sequencing does more than manage risk. It creates value. A project with a contracted offtake and a governance framework that ties capital deployment to demand milestones presents a fundamentally different proposition from one where capital chases market assumptions. The governance layer makes that distinction operational.
Institutional grade governance for infrastructure assets rests on five principles, drawn from the OECD, the IMF’s Public Investment Management Assessment, and emerging practice in data centre joint ventures.
Design at formation, not after deployment. Board composition, reserved matters, decision thresholds, deadlock resolution, and exit provisions must all be settled before capital moves. The Harvard Law School Forum on Corporate Governance has found that governance quality in joint ventures correlates directly with sustained financial performance, risk management, and adaptability. Retrofitting governance once problems surface costs more and works less well than designing it correctly from the start.
Calibrate decision rights to the nature of the decision. Operational decisions belong with management, within guardrails. Strategic decisions, offtaker selection, capital expenditure above defined thresholds, and technology platform changes require board approval with voting thresholds that reflect the partner structure. Constitutional changes, related party transactions, and material litigation require supermajority or unanimous consent.
Accommodate phased development. A data centre campus growing from 50 MW to 500 MW over a decade will pass through fundamentally different commercial stages. Governance must anticipate these transitions and define how decision authority, capital calls, and partner obligations evolve as the project scales.
Meet institutional reporting standards from day one. This means comprehensive investor reporting, performance metrics, capital deployment updates, and regulatory compliance tracking. Governance and administration, as CSC’s analysis emphasised, are not back-office functions. They are central to whether a platform can scale.
Manage conflicts structurally, not through disclosure alone. In data centre joint ventures, conflicts arise when one partner is also the operator, when related party transactions are involved, or when the venture competes with an existing portfolio asset. Protocols for managing these conflicts, including independent review, must be embedded in the governance framework from the outset.
Capital is abundant. Execution capability is scarce. In this environment, governance is increasingly what separates platforms that attract institutional capital from those that cannot.
GCC sovereign wealth funds, European pension funds, and global infrastructure investors are converging on a shared expectation: governance must be designed to the same standard as the physical infrastructure it oversees. A platform where decision mechanics are transparent, capital deployment is sequenced against contracted demand, and risk oversight is independent offers institutional investors something most data centre projects do not: confidence that the asset will be managed to protect value across its full lifecycle.
The sector is entering a phase where governance design will determine which platforms scale and which stall. The GPUs, cooling systems, and power connections are necessary. They are not sufficient. Without the governance layer that determines how decisions are made, how risks are allocated, and how the interests of capital providers, operators, and offtakers are aligned, even a technically excellent facility remains exposed to the quiet erosion that poor governance produces.
The question is not whether governance matters. It is whether the framework being built today will prove resilient across the 20 to 30-year life of the assets it governs. The evidence suggests that most will not, unless governance receives the same rigour currently reserved for power, cooling, and compute.
1. JLL, "2026 Global Data Center Outlook." Link
2. IMF, "How Strong Infrastructure Governance Can End Waste in Public Investment." Link
3. World Bank, "Successful Infrastructure Projects Require Efficient Governance." Link
4. OECD, "Infrastructure Governance Toolkit." Link
5. World Bank, "Overview of the Infrastructure Governance Framework." Link
6. CSC / Lexology, "Data Centers and the Governance Advantage." Link
7. Gibson Dunn, "Joint Venture Governance Best Practices." Link
8. Norton Rose Fulbright, "Joint Ventures Governance Considerations." Link
9. World Bank PPP Legal Resource Center, "Project Company Governance." Link
10. Greenberg Traurig, "Saudi Arabia's Data Centre Expansion." Link
11. Pinsent Masons, "Saudi Arabia's Push to Build Regional Data Centre Hub." Link
12. Reed Smith, "Saudi Arabia's Global AI Hub Law." Link
13. Deloitte, "GCC Sovereign Wealth Funds at the Forefront of Global Expansion." Link
14. Deloitte Middle East, "Gulf SWFs Lead Global Growth." Link
15. Skadden, "Sovereign Wealth Funds and Liberalised Rules Driving Middle Eastern Business." Link
16. Harvard Law School Forum, "Joint Venture Governance Index." Link
17. Harvard Law School Forum, "Recent Trends in Joint Venture Governance." Link
18. WEF, "The State of Play in the Global Data Centre Gold Rush." Link
19. Data Centre Magazine, "Financing the Future, Trends in 2025." Link
20. IMF, "Infrastructure Governance Assessment Framework (PIMA)." Link
21. Clifford Chance, "Data Centre Insights 2025." Link
22. CEPR, "Infrastructure Governance Failures." Link
23. NN Investment Partners / Infrastructure Investor, "Why Good Governance Is Essential." 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.