In the second quarter of 2025, an estimated $98 billion worth of data centre projects in the United States were blocked or delayed by local opposition. That figure exceeds the total disrupted in all prior quarters since 2023 combined. By year’s end, 25 projects had been cancelled outright, moratoriums had been enacted in at least 14 states, and more than 230 environmental organisations had signed a letter to Congress calling for a national freeze on all new construction.
During the same period, Saudi Arabia broke ground on the Hexagon Data Centre (480 MW, Tier IV, $2.7 billion), commenced HUMAIN’s first two AI campuses in Riyadh and Dammam, and published a National Data Centre Strategy targeting 1.5 GW of capacity by 2030. None of these projects encountered significant public opposition.
The gap is worth examining. It does not reflect the absence of constraints in one market and the presence of them in another. It reflects a difference in when and how those constraints are addressed.
The standard model treats political risk as exogenous: an unpredictable shock that arrives after a project is conceived. Insurance products from institutions such as the World Bank’s Multilateral Investment Guarantee Agency (MIGA) are designed for precisely this framing. The assumption is that political risk is something that happens to projects.
The evidence suggests otherwise. A joint framework from the World Economic Forum and Boston Consulting Group on political and regulatory risk in infrastructure found that while such risk manifests at every stage of a project’s lifecycle, the capacity to mitigate it is overwhelmingly concentrated at the front end. Bent Flyvbjerg’s research across hundreds of megaprojects in twenty countries reached a similar conclusion: the systematic underestimation of risks and overestimation of benefits is not accidental but embedded in how projects are originated. His term for it is "design by deception," the practice of circumventing genuine stakeholder engagement to accelerate approval.
The OECD’s governance framework reinforces the point from the institutional side: weak planning disrupts delivery, impedes resource allocation, and erodes the public trust on which long duration infrastructure depends.
Data centres are unusually exposed to this dynamic. A typical campus operates on a 20 to 30 year horizon, consumes large quantities of power and water, generates limited direct employment, and serves customers whose identities and purposes may be opaque to the host community. These characteristics make political goodwill a design input, not a post hoc negotiation. When it is treated as the latter, the results are predictable.
The United States now offers a near laboratory for the study of design stage failure. According to Data Center Watch, opposition groups succeeded in blocking or delaying two out of every three projects they contested in 2025, a 66% strike rate that points to vulnerability more systemic than situational.
The opposition is bipartisan. Among elected officials who publicly opposed data centre projects, roughly 55% were Republican and 45% Democrat. The grievances are consistent across party lines: water consumption, energy grid strain, noise, residential electricity price increases, and insufficient local economic benefit. Polling found that only 44% of Americans would welcome a data centre near their home, a lower approval rate than for gas fired power plants, wind farms, or nuclear reactors.
Legislatures responded accordingly. Lawmakers across all 50 states considered at least 238 data centre related bills in 2025, enacting over 40 in 21 states. Moratoriums proliferated from San Marcos, Texas, to Prince George’s County, Maryland, to Birmingham, Alabama.
The pattern in each case was similar. Sites were selected without mapping the stakeholder landscape. Permitting was pursued as a transactional exercise. Economic benefit proposals failed to address the concerns communities actually held. The failures were not primarily technical or financial. They were failures of governance design.
Europe exhibits the same pattern at lower intensity. Ireland, the Netherlands, and parts of Germany have enacted or are considering moratoriums, driven primarily by grid capacity limits and environmental concerns. The lesson is consistent across jurisdictions: identical capacity, deployed under different governance frameworks, produces different political outcomes.
The Kingdom of Saudi Arabia offers the most developed counterexample, not because it is free from constraints, but because its approach addresses them at inception rather than at permitting.
The civic and institutional contexts of the two countries are plainly different. Saudi Arabia is a centralised monarchy; the United States is a federal democracy with dispersed permitting authority. But the analytical point holds across systems: projects that embed stakeholder alignment and regulatory coordination into their design are more durable than those that do not, regardless of the political environment in which they operate.
Saudi data centre development is a sovereign priority integrated into Vision 2030. Sixty six of the Vision’s 99 goals are directly linked to data and AI, according to SDAIA. The institutional architecture is designed to match. SDAIA operates as both regulator and strategic driver, overseeing the Personal Data Protection Law, AI governance standards, and the national data centre programme simultaneously. The Ministry of Communications and Information Technology (MCIT) co authored the National Data Centre Strategy. The Communications, Space and Technology Commission (CST) supervises the Cloud Computing Special Economic Zone.
This coordination reduces the regulatory fragmentation that creates political risk elsewhere. In the United States, a single project may require approvals from local planning commissions, state environmental agencies, federal energy regulators, and utility companies, each operating on different timelines with different mandates. In Saudi Arabia, these functions are consolidated under a national programme.
HUMAIN, created in May 2025 and owned by the Public Investment Fund, is an operating company, not a holding vehicle. It is building end to end AI capabilities across infrastructure, cloud, and models.
Its approach to data centre development illustrates what design stage discipline looks like at scale. HUMAIN does not originate sites and then seek regulators and offtakers. It operates within a sovereign framework where demand, regulation, and capital are sequenced before construction begins. Its partnerships (AirTrunk and Blackstone for a $3 billion initial campus, stc Group’s center3 for up to 1 GW of AI workloads, Google Cloud for a $10 billion AI hub, AMD for $10 billion and 500 MW of compute capacity) are structured as demand validated deployments, not speculative builds.
The PIF’s trajectory reinforces this. As the world’s most active sovereign wealth fund in 2025, deploying $36.2 billion over the year, it is now preparing a revised 2026 to 2030 strategy that prioritises AI alongside industrial manufacturing, logistics, and tourism.
The Cloud Computing Special Economic Zone, launched in April 2023, is the operational expression of regulatory integration. Electricity is priced at $0.05 per kilowatt hour, roughly half the US average. Customs duties on capital equipment are deferred or eliminated. Tax treatment follows OECD principles. Critically, the zone is virtual: companies established within it can build anywhere in the Kingdom, avoiding the geographic concentration that has triggered opposition in Northern Virginia, Dublin, and Amsterdam.
The CCSEZ targets 30% of the Kingdom’s ICT spending by 2030 and aims to attract $20 billion in investment. Its purpose is to create an environment in which data centre development is supported by default rather than negotiated project by project.
The Kingdom’s power market (93 GW installed, projected to reach 123 GW by 2030, with a target of 58.7 GW in renewables) and land economics ($10 to $50 per square metre versus $150 to $600 in Northern Virginia) provide a cost foundation that most competing markets cannot match. The grid expansion trajectory addresses the energy constraint proactively, unlike in the United States, where data centres are projected to consume 12% of national electricity by 2028 and where that consumption is already fuelling opposition.
These advantages, however, coexist with material constraints. The Kingdom faces a roughly 50% hiring gap for AI and data centre specialists. Cooling costs in desert environments can account for a significant share of total operating expenditure. The regulatory framework is advancing rapidly, which also means it has not yet fully stabilised, creating uncertainty for operators accustomed to more mature regimes. And the PIF’s own fiscal recalibration, with capital spending adjustments of up to 15% anticipated as it revises its investment strategy, confirms that sovereign backing has limits. Projects that do not meet economic viability thresholds will be deprioritised.
Acknowledging these constraints is not a caveat to the design stage thesis. It is the thesis. These are precisely the factors that constraint mapping must identify and address at inception.
The research on infrastructure governance identifies four design decisions that shape a project’s political durability over its full lifecycle:
The academic literature is unambiguous: weak stakeholder engagement is among the most common drivers of project failure. The question of where to build must be answered by technical, commercial, and political criteria together. HUMAIN’s selection of Riyadh and Dammam for its first campuses was made within the framework of Vision 2030’s regional development priorities, aligning site decisions with sovereign objectives from inception.
Research published in the Journal of Management in Engineering found that governance quality correlates with project performance only when stakeholder management is functioning at a medium or higher level. In Saudi Arabia, this is achieved at the national level: the PIF allocates 80% of its $1.15 trillion in assets domestically, and Vision 2030 targets training over 100,000 digital infrastructure specialists by 2030, integrating employment and skills development into the same programme that delivers capacity.
Compliance is reactive: build the project, then seek approval. Integration is proactive: the regulatory framework is an input to design, shaping site selection, power strategy, and environmental commitments from the outset. The CCSEZ is the jurisdictional expression of this principle. Developers enter a pre established environment rather than negotiating bespoke terms.
Deploying capital against speculative capacity is one of the most reliable generators of political risk in infrastructure. It exposes every stakeholder to risk that disciplined sequencing would eliminate. The principle that capital follows contracted demand is not only financial prudence; it is a political risk control. HUMAIN’s partnership structure, where each counterparty committed against defined capacity and workload requirements, applies this discipline at sovereign scale.
These four principles are not exclusive to sovereign backed entities. Any platform that combines infrastructure execution, demand origination, and capital sequencing within a single operating model can achieve comparable durability, provided the discipline is applied at inception. What matters is the methodology, not the ownership structure.
If political risk is created at the design stage, then risk mitigation must also move there. This requires something more expansive than conventional due diligence.
Standard due diligence asks whether a project can be built. Constraint mapping asks whether it can survive. The distinction is not semantic. The first is answered by engineering and permitting analysis. The second requires assessment of political dynamics, regulatory trajectory, energy grid evolution, sovereign priorities, and the alignment between the project’s value proposition and the host jurisdiction’s long term objectives.
In Saudi Arabia, constraint mapping means understanding not only grid interconnection and power procurement, but also the political economy of Vision 2030, the PIF’s evolving investment criteria, SDAIA’s regulatory trajectory, and the Kingdom’s competitive position relative to the UAE and Qatar. In European markets, it means mapping renewable power availability, data sovereignty requirements, and grid capacity against emerging corridors in the Nordics, France, and Iberia that offer alternatives to saturated hubs. In the United States, 238 bills across all 50 states in a single year signal a regulatory environment that is not hostile to data centres per se, but is recalibrating the terms on which they are permitted to operate.
A project that maps these constraints at inception has a different risk profile from one that does not, even if both sit in the same jurisdiction on the same grid.
The conventional approach to political risk in infrastructure investment applies it as a discount factor. A project in a jurisdiction perceived as higher risk receives a higher hurdle rate. Insurance products or government guarantees are layered in. This approach has the merit of simplicity, but it rests on a flawed assumption: that political risk is a fixed attribute of the jurisdiction.
It is not. It is a variable attribute of the project’s design.
Two data centres in the same market can have fundamentally different political risk profiles depending on how they were originated, who was consulted during design, whether demand was contracted before capital was committed, and whether the governance framework delivers visible benefit to the host community. A facility conceived in alignment with sovereign priorities and phased to match contracted demand is a different asset class from one sited by an optimisation algorithm and financed against speculative capacity.
The Kingdom of Saudi Arabia’s ecosystem illustrates this at scale. The convergence of PIF capital, HUMAIN’s operational capability, SDAIA regulatory authority, CCSEZ fiscal architecture, and national power grid expansion creates conditions in which data centre projects are designed for political durability. Capital follows demand. Regulation precedes construction. Sovereign alignment is an input, not an afterthought.
The MIGA framework covers expropriation, breach of contract, currency restrictions, and civil disturbance. The most common political risks facing data centre investments today (community opposition, regulatory tightening, energy policy shifts, failures of sovereign alignment) are not covered, because they are not insurable events. They are design failures. The only effective hedge is better design.
The evidence from 2025 is clear in its implications: the primary source of political risk in data centre infrastructure is not geopolitical instability, regulatory hostility, or community irrationality. It is the set of decisions made, or not made, at the design stage.
The next phase will test whether design stage discipline scales. Saudi Arabia’s trajectory extends from 1.5 GW by 2030 to 6 GW by 2034, a fourfold expansion requiring sustained coordination across sovereign entities, international operators, hyperscale offtakers, and institutional capital. The PIF’s fiscal recalibration, HUMAIN’s operational growth, evolving chip export regulations, and competitive dynamics among GCC states will all exert pressure on the frameworks that have delivered durability thus far. If the methodology holds at scale, the Kingdom will have established that political risk management is a repeatable discipline. If it does not, the failures will indicate exactly where design was insufficient.
For institutional capital evaluating data centre opportunities, the operative question is not which jurisdictions carry political risk. All of them do. The question is whether the project under consideration has addressed that risk at the only stage where it can be effectively controlled.
1. Data Center Watch, Q2 2025 Report on Data Center Opposition (2025) · datacenterwatch.org
2. Data Center Watch, Report on Data Center Opposition (2025) · datacenterwatch.org
3. Heatmap, Local Opposition to Data Centers Is Surging (2026) · heatmap.news
4. Truthout, As Towns and Cities Fight Off Data Centers, Calls Grow for a National Moratorium (2025) · truthout.org
5. NPR, Environmental Groups Call for a Moratorium on Data Center Construction (2025) · npr.org
6. World Economic Forum / BCG, Strategic Infrastructure: Mitigation of Political & Regulatory Risk (2015) · weforum.org
7. Flyvbjerg, B., Megaprojects and Risk: An Anatomy of Ambition, Cambridge University Press (2003)
8. OECD, Infrastructure Governance · oecd.org
9. Greenberg Traurig, Saudi Arabia’s Data Centre Expansion: Regulatory Framework and Strategic Considerations (2026) · gtlaw.com
10. Gulf News, Saudi Arabia Unveils World’s Largest Government Data Centre in $2.7 Billion Digital Push (2026) · gulfnews.com
11. Arabian Business, Saudi Arabia Launches $2.7bn Digital Infrastructure Push (2026) · arabianbusiness.com
12. CNBC, Saudi AI Firm Humain Is Pouring Billions into Data Centers (2025) · cnbc.com
13. CNN, Saudi Arabia Is Making a Massive Bet on Becoming a Global AI Powerhouse (2025) · cnn.com
14. Blackstone, HUMAIN and Blackstone Backed AirTrunk Announce Partnership (2025) · blackstone.com
15. PwC Middle East, Unlocking the Data Centre Opportunity in the Middle East (2025) · pwc.com
16. US International Trade Administration, Saudi Arabia ICT New Data Center Strategy (2025) · trade.gov
17. CST, Cloud Computing Special Economic Zone · cst.gov.sa
18. Arab News, Riyadh’s Cloud Computing Economic Zone a Game Changer (2024) · arabnews.com
19. Middle East Briefing, Saudi Arabia’s Evolving Regulatory Framework for the Technology and Data Economy (2025) · middleeastbriefing.com
20. Crowell & Moring, The Middle East’s Big Bet on Artificial Intelligence and Data Security (2026) · crowell.com
21. Gulf News, Saudi Arabia’s PIF Named World’s Most Active SWF in 2025 (2026) · gulfnews.com
22. AGBI, Saudi to Test Revised PIF Strategy as Giga Project Funding Tightens (2026) · agbi.com
23. Khan, A. et al., “Improving the Performance of Public Sector Infrastructure Projects,” Journal of Management in Engineering, Vol 37 (2021) · ascelibrary.orgME.1943-5479.0000886)
24. Prebanić, K.R. and Vukomanović, M., “Exploring Stakeholder Engagement Process,” Buildings, 13(7) (2023) · mdpi.com
25. MIGA / World Bank Group, Political Risk Insurance · miga.org
26. Data Center Knowledge, Calls for US Data Center Freeze Grow (2025) · datacenterknowledge.com
27. Tech Policy Press, Local Governments Are Learning How to Negotiate with Data Center Developers (2025) · techpolicy.press
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.