Powering AI: Energy, Land and the Real Bottlenecks in Gulf Data-Centre Investment

Introduction

The story of the AI data-centre boom is usually told as a demand story: artificial intelligence requires vast computing capacity, the demand is enormous and growing, and so capacity must be built. But the demand, however large, is not the constraint. The constraint is supply, and the supply of data-centre capacity is limited not by capital or by willingness to build but by physical bottlenecks, chiefly power, then suitable land, cooling and connectivity, that take time and effort to overcome. These bottlenecks determine which projects can actually be delivered, how quickly, and at what cost, and therefore which investors earn the returns the demand implies. An investor that understands the demand but not the bottlenecks will misjudge the opportunity, because it is in the bottlenecks, not the demand, that the returns are won and lost.

This paper sets out the real bottlenecks behind the Gulf data-centre buildout and how an investor should price and invest around them. It is written for the infrastructure investor, the developer and the lender who must assess not whether demand exists, which it does, but whether a given project can secure the power, land and connectivity it needs, on the timeline it assumes, at the cost it projects. Its purpose is to move the investor’s attention from the demand, which is well understood, to the supply constraints, which are less so and which actually determine outcomes. The central message is that power, in both its availability and the speed of connecting it, is the dominant bottleneck; that the value of a project lies substantially in its time-to-power; and that the investor’s returns depend on securing the binding inputs early and pricing the lead times that less careful participants underestimate.

The companion papers in this cluster make the strategic case for the buildout and set out the underwriting of a specific asset; this paper examines the physical constraints that both must contend with. Where the strategic paper identifies power and offtake as the dominant value drivers, this paper develops the power and land constraints in depth, and where the underwriting paper treats power security as a gating item, this paper explains why it gates and how to assess it. The three together give an investor the demand, the asset and the constraints, and this paper supplies the constraints that the others take as given.

The Gulf context is, on the power dimension, unusually favourable, which is part of why the region is building so aggressively: energy is abundant and inexpensive, and increasingly complemented by large-scale solar. But abundance at the national level does not eliminate the project-level bottleneck, which is the speed and certainty of connecting firm power to a specific site on the project’s timeline, and which depends on grid capacity, substation availability and the permitting and construction of connections. The Gulf’s structural energy advantage eases but does not remove this constraint, and the investor must assess it project by project rather than assuming the national abundance flows automatically to its site.

Results And Discussion

This section presents the framework in the order of the propositions: the ranking of bottlenecks (Proposition 1); the power solutions, lead times and costs (Propositions 2 and 3); the site-suitability factors (Proposition 4); the time-to-power; and the impact of power constraints (Proposition 5).

A useful way to internalise the inverted structure is to ask, of any data-centre project, what would actually stop it. It is rarely a lack of customers, given the demand; it is far more often that power cannot be secured, that the site cannot be connected, that the permits do not come, or that the timeline stretches until the opportunity passes. These are the things that stop projects, and they are all supply constraints. An investor that habitually asks what would stop this project, rather than will there be demand for it, is led directly to the bottlenecks, and to the diligence and securing actions that address them. This question, what would actually stop it, is a simple but powerful discipline for keeping the investor’s attention on the constraints that determine outcomes, and it naturally produces the supply-focused analysis the framework recommends, because the honest answers are almost all about supply.

4.0a Why Demand Is Not the Constraint

Before ranking the bottlenecks, it is worth establishing the premise that distinguishes this analysis: that demand, for once, is not the binding constraint. In most investment, demand uncertainty dominates, the central question is whether customers will want the product, and supply is assumed to follow demand. The AI data-centre boom inverts this. The demand for computing capacity is, for the foreseeable horizon, expected to exceed the capacity the industry can build, so the question of whether there will be demand for a well-located, well-powered data centre is, within reason, settled in the affirmative. What is not settled is whether a given project can secure the power, land and connectivity to be built and to deliver capacity, and when. This inversion, demand abundant, supply constrained, is why the analysis can take demand largely as given and concentrate on the supply bottlenecks, and it is a genuinely unusual feature that investors accustomed to demand-driven analysis must consciously adopt. The investor that keeps worrying about demand while neglecting supply has not adjusted to the inverted structure of this particular opportunity.

This premise should be held with appropriate qualification, as the companion strategic paper notes: demand could disappoint if AI adoption stalls or computing efficiency leaps, and the breadth of demand drivers, not only AI, cushions but does not eliminate that risk. But for the purpose of analysing the bottlenecks, the relevant point is that for any plausible demand path short of a broad collapse, the supply constraints bind before the demand does, so the marginal project’s fate is decided by whether it can be powered and built, not by whether its capacity will be wanted. The investor should therefore size its overall exposure to the theme with the demand risk in mind, as the strategic paper counsels, but should analyse each specific project through the supply bottlenecks, because that is where the individual project succeeds or fails. Holding both lenses, demand risk at the portfolio level, supply constraints at the project level, is the complete posture, and this paper supplies the project-level supply lens.

Ranking the Bottlenecks

The analysis begins by identifying what actually constrains projects. Figure 1 ranks the bottlenecks.

Figure 1. The Real Bottlenecks in Gulf Data-Centre Buildout

Indicative; power, in availability and connection speed, constrains the most projects, ahead of land, cooling and connectivity.

The figure supports Proposition 1. Power, taken as both the availability of capacity and the speed of connecting it, constrains the largest share of projects, ahead of suitable land, cooling and water, connectivity, and skilled labour. The dominance of power reflects its nature: a data centre is, in essence, a facility for converting large amounts of electricity into computing, so power is its defining input, and the quantity required, rising as AI workloads grow denser, strains the grids and the connection processes that must deliver it. The other bottlenecks matter, and can bind on particular projects, but power is the one most likely to be the binding constraint, and it is therefore the one on which the investor should concentrate its attention. The practical lesson is that an investor assessing a project should ask first and most rigorously about power, because that is where the project is most likely to be constrained, and a project with a clear, secured power solution has overcome the bottleneck most likely to defeat it.

Implementation Considerations

Translating the framework into practice involves securing power first, selecting sites for infrastructure access, pricing lead times realistically, and building optionality on scarce inputs.

A foundational sequencing point, parallel to the other papers in this cluster, is that the power question must be settled before, not after, the capital is committed and the site acquired, because the investor’s flexibility is greatest before it has sunk costs into a particular project. Before committing, the investor can choose among sites and power solutions, walk away from a project whose power cannot be secured, and negotiate from a position of choice; after committing to a site and incurring costs, it is anchored to that project and its power problem, with far less flexibility to respond if the power proves hard to secure. The discipline is therefore to resolve the power question, to secure firm power or a clear, evidenced path to it, as a condition of committing, so that the investor never finds itself holding a site it cannot power. This sequencing, power before commitment, is the single most important procedural discipline the framework implies, because it preserves the investor’s ability to avoid the worst outcome, a committed, unpowered project, which is the outcome the bottleneck analysis most warns against.

Securing Power First

The foundational discipline is to secure firm power before committing to a project, treating an unsecured power position as a reason to wait rather than a risk to accept. This means obtaining a contracted, firm power supply on the project’s timeline, including for planned expansion, and understanding the lead time and certainty of the connection. An investor that secures power first has overcome the dominant bottleneck and can proceed with confidence; one that commits before securing power exposes itself to the compounding delay, cost and shortfall that power constraints inflict. Power should be the first question and the first secured item, not a matter resolved along the way.

A practical refinement on site selection is to assess not only a site’s current infrastructure access but its trajectory, because grid and fibre networks evolve. A site that is currently power-proximate may face rising competition for the same grid capacity as other developments cluster around it, while a site currently less connected may sit on a planned grid expansion or fibre route that will improve its access. The investor should therefore assess the forward path of the infrastructure around a site, the planned grid investments, the committed fibre builds, the other developments competing for the same capacity, rather than only its present state, because the power and connectivity available when the project is built and operating matter more than those available today. This forward assessment requires engagement with the utilities and planners who shape the infrastructure’s trajectory, reinforcing the value of the early relationships the analysis recommends, and it guards against the error of selecting a site for a current access that competition or congestion may erode by the time the project needs it.

Selecting Sites for Infrastructure Access

Because land is valued for its access to power, fibre and water, the investor should select sites primarily for their infrastructure proximity rather than their cost or size. A site adjacent to spare grid capacity and a fibre route, with water for cooling and clear permitting, is worth more for a data centre than a cheaper, larger site lacking these, because it can be powered and connected quickly. The investor should assess each candidate site for its access to the binding inputs and weight that access far above land price, recognising that the apparent saving on a poorly-connected site is illusory if the site cannot be powered on the timeline the project needs.

Concluding Comments

The Gulf data-centre boom is a demand story with a supply problem, and it is the supply problem, the physical bottlenecks of power, land, cooling and connectivity, that determines which projects succeed and what returns investors earn. This paper has set out the real bottlenecks, shown that power dominates, and explained why time-to-power drives value and how an investor should invest around the constraints.

The evidence and analysis support five conclusions. First, power, in availability and connection speed, is the dominant bottleneck. Second, the value of a project lies substantially in its time-to-power, so a grid-ready site commands a premium. Third, power cost varies by source, and the Gulf’s cheap conventional energy and abundant solar are a structural advantage, subject to securing firm, timely supply. Fourth, land is valued for its proximity to power, fibre and water, not for itself. Fifth, the investor’s returns depend on securing the binding inputs early and pricing the lead times others underestimate.

Before turning to the limitations, it is worth placing this paper within its cluster. The strategic companion paper establishes that the Gulf data-centre opportunity is large and that power and offtake are its dominant value drivers; the underwriting companion treats power security as a gating item in lending against a specific asset; and this paper explains why power gates, how the constraint behaves, and how an investor should invest around it. Together they take an investor from the strategic case, through the underwriting of an asset, to the physical constraints that determine whether any asset can be built. This paper’s contribution is to make the supply constraints, which the others take as given, the explicit object of analysis, so that the investor neither assumes the supply will follow the demand nor underestimates the lead times that ration the opportunity to those who solve them. The bottlenecks are the part of the data-centre story least discussed and most decisive, and this paper aims to give them the central place the returns show they deserve.