Offtake Is Not a Contract. It Is a System

Global investment in data centre infrastructure will exceed $3 trillion over the next five years, according to Moody’s 2026 outlook. The five largest US hyperscalers spent more than $405 billion in 2025. Analyst consensus projects aggregate hyperscaler capital expenditure of approximately $600 billion in 2026, a further 36% increase, according to CreditSights and Goldman Sachs Research.

Across the GCC, the market is forecast to grow from $3.48 billion to $9.49 billion by 2030, a compound annual growth rate of 18.2%, according to ResearchAndMarkets. Skadden has called it a "generational super cycle." In this environment, "offtake" has become shorthand for demand security. Secure the contract, and the rest follows: capital, construction, returns. This framing is materially incomplete. For institutional investors directing capital into AI infrastructure, particularly across the Gulf, it invites precisely the kind of risk it claims to resolve.

An offtake agreement does not extinguish demand risk. It restructures it. The contract initiates a relationship between counterparties, incentive systems, regulatory regimes, and operational realities that together determine whether contracted demand produces realised revenue. Most data centre project failures do not occur because a contract was never signed. They occur because what was signed could not be sustained.
‍

The contract is a surface

A typical hyperscale offtake agreement looks, on paper, like an exceptional asset. As Lexology’s analysis of data centre financings notes, lenders and equity holders underwrite on the creditworthiness of the tenant, the quality of projected cash flows, and the risk profile of the lease. A 15 to 20 year commitment from a hyperscaler offers investment grade credit and long duration revenue visibility.

The structural dynamics beneath that surface tell a different story. DLA Piper’s 2025 assessment of hyperscale lease structures identifies a growing tension between long lease terms and tenants’ expanding termination rights. In several cases in 2025, tenants exercised those rights or walked away from major commitments, leaving developers with vacant facilities and lenders with concentrated exposure.

The asymmetry runs deeper. Ropes and Gray observe that hyperscalers now routinely negotiate restrictions on asset transfers, constraints on financing counterparties, and exit provisions that shift residual risk to the landlord. The counterparty’s credit quality is not in doubt. The power balance within the contract is.

Moody’s reinforces the point: pre leasing to hyperscalers limits vacancy risk but simultaneously increases counterparty concentration risk. When a single tenant occupies an entire campus, renewal is binary. A non renewal does not erode revenue gradually. It eliminates it. Infrastructure lenders increasingly reflect this in their underwriting, scrutinising termination provisions and tenant concentration when setting debt service coverage ratio thresholds and lending covenants.

Offtake, then, is not a document that resolves demand risk. It is a system of interlocking conditions that must hold together over years of operational delivery.‍

‍

Five layers beneath the signature

Demand security in AI data centre infrastructure rests on five distinct layers. Each must function for contracted offtake to produce the returns that investors underwrite.

1. ‍Counterparty alignment. The five largest hyperscalers account for roughly 70% of global cloud infrastructure spending and control approximately 60% of installed hyperscale capacity, according to analysis drawing on Synergy Research Group data. This concentration creates gravitational effects. In tight markets, hyperscalers benefit from scarcity. In loose ones, they dictate terms. Assessing whether a 15 year commitment will be honoured in substance requires understanding the counterparty’s capital expenditure trajectory, geographic priorities, and technology roadmap, not merely its balance sheet.
   ‍
2. ‍Technical specification integrity. AI workloads evolve faster than the facilities built to serve them. Power densities have risen from 10 to 15 kW per rack in traditional enterprise configurations to 80, 150, and in some deployments over 200 kW per rack. McKinsey and Synergy Research Group data confirm that the technical requirements embedded at the point of signature may diverge materially from what the offtaker needs 24 or 36 months later.
   ‍
   The risk is not that every facility becomes obsolete with each chip generation. Well designed facilities with sufficient power headroom and flexible cooling architectures accommodate multiple GPU cycles. The risk is that facilities built to minimum specifications cannot adapt. When that happens, the contract’s economic value degrades regardless of its legal enforceability. Moody’s 2026 outlook notes that advances in chip design and cooling architecture increase obsolescence risk for newer data centres even as they are being commissioned.
   ‍
3. ‍Power and grid certainty. Power availability, grid capacity, and cooling efficiency define the binding constraints on data centre development in the GCC, as MEED’s 2026 market analysis details. Saudi Arabia’s Data Centre Services Regulations, introduced alongside the Kingdom’s target of approximately 1.5 GW of capacity by 2030, explicitly recognise that rapid expansion must be managed within grid constraints. Greenberg Traurig’s regulatory analysis flags that clustering of large facilities may stress local networks, that supply chain constraints in specialised equipment can affect project schedules, and that siting decisions must align with planned grid reinforcements.
   
   Globally, wait times for new power connections exceed five years in some primary markets. In 2025 and 2026, AWS and Meta publicly paused construction on billion dollar facilities to reassess whether [[promised grid connections would materialise]]. A signed offtake agreement is worth little if the facility cannot be energised on schedule.
   ‍
4. ‍Regulatory and sovereign alignment. Data sovereignty mandates in the GCC create structural demand for domestic infrastructure. Saudi Arabia’s data localisation requirements, its Cloud Computing Special Economic Zone (projected to account for 30% of the Kingdom’s ICT spending by 2030, according to [[PwC]], and the UAE’s comparable frameworks compel multinational corporations to host sensitive workloads locally.
   
   Addleshaw Goddard notes that the region’s regulatory architecture has been designed to accelerate development while enforcing data residency. But regulatory regimes evolve. An offtake structured without accounting for emerging ESG, sustainability, and permitting requirements risks misalignment over the life of the asset. The Communications, Space, and Technology Commission (CST) is raising standards for project qualification in the Kingdom, favouring operators with established compliance credentials and certification histories.There is also an external dimension. Access to advanced AI accelerators, including NVIDIA’s latest generation GPUs, operates within the framework of US Bureau of Industry and Security (BIS) export licensing.
   
   Sovereign AI programmes across the Gulf, including HUMAIN in Saudi Arabia, depend on continued access to these supply chains. An offtake structured around a specific GPU architecture carries an implicit assumption about the geopolitical conditions that permit procurement of that hardware. A shift in US export control policy could alter the technical and economic basis of a contract without either party breaching its terms.
   
   The EU AI Act’s evolving compliance requirements add a further layer for GCC operators serving European workloads, a consideration that will become more material as cross jurisdictional data flows intensify.
   ‍
5. ‍Operational execution. Moody’s warns it expects more operational problems as the ranks of newer, less experienced operators expand. In the current supply constrained environment, tenants accept facilities when available rather than exercising termination rights. That forbearance will not last.
   
   Once supply and demand reach better balance, the operational quality of the facility, uptime, power usage effectiveness, cooling performance, maintenance responsiveness, and service level adherence, becomes the determinant of whether offtake holds.

‍

Where speculation concentrates

The risk of treating offtake as a contract rather than a system is most visible in the broader pattern of speculative capacity development. As of June 2025, Data Center Watch reported more than 36 projects worth $162 billion either blocked or significantly delayed. In the United States alone, $64 billion in projects stalled over two years because of local opposition, grid strain, and permitting failures across Virginia, Arizona, California, and elsewhere.
‍

The scale of the mismatch is notable. The CEO of Constellation Energy warned investors in early 2025 that utility projections for data centre power demand in just three US markets exceeded credible forecasts for the entire country. The Institute for Energy Economics and Financial Analysis found that utility forecasts in four states overstated actual industry trends by a factor of four.
‍

Ares Management’s co president Kipp deVeer stated the implication directly at the Greenwich Economic Forum: when this much capacity comes online, some of it will prove marginal. These cycles, he noted, tend to produce overbuilds in certain markets.
‍

The risk is not evenly distributed. Well structured projects with phased power procurement, diversified counterparty exposure, and technical flexibility face a materially different profile than single counterparty, single jurisdiction developments built against speculative demand forecasts. The concentration of failure is in the latter: projects that treated a letter of intent as sufficient proof of demand without verifying the operational, regulatory, and infrastructure conditions required to sustain it.
‍

Many of these failure patterns cluster in Western markets, where community opposition, fragmented permitting, deregulated grids, and contested zoning create execution friction. These dynamics are less pronounced in the GCC, where centralised planning, sovereign backing, streamlined regulatory approval, and coordinated infrastructure investment reduce several categories of execution risk.

That does not eliminate the need for disciplined offtake structuring in the Gulf. It means the nature of the risks differs. In the GCC, the primary execution challenges are grid integration sequencing, supply chain coordination for specialised equipment, and the pace at which announced capacity targets translate into operational infrastructure, rather than the permitting and community dynamics that dominate Western markets.
‍

Sequencing as discipline

The implication for institutional capital is that the order in which things happen matters as much as whether they happen at all.

Capital deployed before demand is contracted is, in most infrastructure contexts, speculative. But capital deployed after a contract is signed, yet before power, technical alignment, and regulatory compliance are confirmed, is nearly as exposed. The contract creates a liability without certainty that the facility can perform to the standard required to retain the tenant.

The GCC is better positioned than many markets to manage these risks. Saudi Arabia’s grid electricity is priced at approximately $0.05 per kWh, compared with $0.10 to $0.15 per kWh in primary US data centre markets, according to Mordor Intelligence and the Gulf Data Centre Alliance. The region offers abundant land, strong government backing, and regulatory frameworks designed to attract sovereign and hyperscale capacity.

Riyadh is the Kingdom’s primary data centre hub, benefiting from proximity to government institutions, enterprise demand, and expanding fibre infrastructure. The Saudi Ministry of Communications and Information Technology has outlined investments exceeding $18 billion in data centres and renewable energy, targeting over 1,300 megawatts of capacity before 2030.

The Kingdom’s renewable energy pipeline now exceeds 58 GW of solar and wind projects either operational, under development, or tendered, according to AGSI analysis, reinforcing the long term cost advantage for power intensive infrastructure. HUMAIN, a subsidiary of the Public Investment Fund, has stated its objective of positioning the Kingdom among the top three nations globally in AI infrastructure.

Stargate UAE, the 1 GW AI compute cluster being developed by a consortium including G42, OpenAI, Oracle, NVIDIA, Cisco, and SoftBank under the US UAE AI Acceleration Partnership, and Qatar’s Digital Agenda 2030 signal a region moving with both strategic intent and institutional capital.

But intent, even when backed by sovereign capital and active project pipelines, must be matched by execution discipline. PwC projects that Middle Eastern capacity will triple from 1 GW in 2025 to 3.3 GW over the next five years. Ambition at this scale introduces its own constraints. Greenberg Traurig flags that clustering may stress local networks, that specialised hardware has long lead times, and that siting decisions must align with planned grid reinforcements. The sequencing discipline required to navigate these constraints separates projects that deliver from projects that stall.
‍

Systems, not signatures

Treating offtake as a system rather than a contract changes how projects are structured, assessed, and governed.

It means evaluating a counterparty not for credit quality alone, but for strategic alignment with the facility’s technical roadmap and geographic position. It means validating power procurement, grid interconnection, and cooling architecture against offtaker requirements before capital commitment, not after. It means structuring agreements that anticipate the evolution of AI workloads, regulatory standards, and ESG expectations over the life of the asset. And it means building operational capacity that can sustain the relationship through cycles of supply and demand rebalancing.

In rapidly scaling markets where regulatory infrastructure is still maturing, where grid capacity is expanding in parallel with facility construction, and where announced capacity targets are still being converted into operational assets, these disciplines are not optional. In the GCC, the opportunity is substantial. So is the gap between a signed agreement and a performing asset.

For investors, operators, and capital partners, the question is not whether offtake has been secured. It is whether the conditions required to sustain it, counterparty incentives, technical alignment, power certainty, regulatory compliance, and operational capability, have been systematically verified and sequenced.

Offtake is not a milestone. It is a discipline.

Sources

1. Moody’s, “2026 Data Center Outlook,” January 2026. Link

2. ResearchAndMarkets, “GCC Data Center Market Investment Analysis Report 2025--2030,” March 2025. Link

3. DLA Piper, “Navigating Hyperscale Lease Terms and Termination Rights,” October 2025. Link

4. Ropes and Gray, “Demand for Hyperscale Data Centres,” June 2025. Link

5. Lexology, “Key Insights in Data Centre Financings,” July 2024. Link

6. Skadden, “Hyperscaler Data Centers: Financing Solutions for Large Scale Projects,” September 2025. Link

7. Global Data Center Hub, “The Hidden Risk Inside the Hyperscale Boom,” November 2025. Link

8. IEEFA, “Risk of AI Driven Overbuilt Infrastructure Is Real,” 2025. Link

9. Data Center Watch, “$64 Billion of Data Center Projects Blocked or Delayed,” March 2025. Link

10. Bloomberg, “Data Center Boom Brings Risks of Overbuilding, Ares’ deVeer Says,” October 2025. Link

11. Data Center Knowledge, “AI First Hyperscalers: 2026’s Sprint Meets the Power Bottleneck,” February 2026. Link

12. Moody’s, “Data Centers: Managing Risk Amid a Market Boom,” January 2026. Link

13. Data Center Knowledge, “Did Hyperscalers Solve the Power Problem in 2025,” December 2025. Link

14. CSG Talent, “Data Centre Grid Challenges: Powering AI Infrastructure at Scale,” 2025. Link

15. Greenberg Traurig, “Saudi Arabia’s Data Centre Expansion: Regulatory Framework and Strategic Considerations,” January 2026. Link

16. PwC, “Unlocking the Data Centre Opportunity in the Middle East,” 2025. Link

17. MEED, “The GCC Data Centre Projects Market 2026.” Link

18. Addleshaw Goddard, “The Future of Data Centres in the Gulf Cooperation Council,” 2025. Link

19. Gulf Data Centre Alliance, “5 Things to Know About the GCC Data Centre Market.” Link

20. Mordor Intelligence, “Saudi Arabia Data Center Power Market Size, Trends, Share and Global Report 2030.” Link

21. Freshfields, “The Digital Gold Rush: Investment in Data Centres and Related Legal Issues,” 2025. Link

22. Goldman Sachs Research, “Why AI Companies May Invest More than $500 Billion in 2026,” December 2025. ; CreditSights, “Technology: Hyperscaler Capex 2026 Estimates,” November 2025. [[CreditSights]](https://know.creditsights.com/insights/technology-hyperscaler-capex-2026-estimates/) Link

23. Saudi Ministry of Communications and Information Technology, “Saudi Arabia Expands Plan to Develop Digital Infrastructure to Build and Enable Mega Data Centers.” Link

24. AGSI, “Saudi Arabia Ramps Up Renewables,” September 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.