The Saudi Data Center Footprint in 2026

Saudi Arabia’s installed and announced AI-grade data center capacity now exceeds 6.6 gigawatts of IT load across roughly two dozen distinct campuses at various stages of permitting, civil works, fit-out, and operation. That figure puts the Kingdom on a path that, if delivered against schedule, would place it inside the global top-five sovereign data-center markets by the end of the decade — behind the United States, China, and the EU as a bloc, and competitive with the United Arab Emirates and India. The trajectory is steep enough that the relevant analytical question is not whether Saudi Arabia will materially scale its compute footprint but how much of the announced pipeline actually energizes on the disclosed timetables.

Saudi infrastructure is not a single market. It is four overlapping clusters with distinct economics, sovereignty postures, and customer profiles. The Riyadh metropolitan cluster anchors enterprise, government, and SDAIA-adjacent workloads. The Eastern Province cluster around Dammam and Jubail is the cheap-power, heavy-industry profile. NEOM and the broader northwest is the moonshot, anchored on DataVolt and the NEOM Tech & Digital Company. The Red Sea and Western corridor — Jeddah, Yanbu, the King Abdullah Economic City — handles connectivity, content delivery, and a growing share of hyperscaler edge. Each cluster has its own permitting environment, its own grid posture, and its own water economics.

Anchor Facilities

Humain’s consolidated campus pipeline is the largest single program inside Kingdom borders, with internal capacity guidance in the 1.9 GW zone over the medium term. The first operational sites are the Riyadh-metro flagship campuses, configured around NVIDIA GB300 NVL72 reference designs with liquid-to-the-chip cooling and direct-current distribution to the rack. Site selection has prioritized proximity to the SDAIA technical campus and to the King Abdulaziz City for Science and Technology (KACST), so that the Saudi national HPC asset Shaheen III — the HPE Cray EX system at KAUST that delivered roughly 120 petaflops of double-precision performance at commissioning — sits inside a coherent national compute fabric rather than as an island.

DataVolt’s NEOM program is the largest single announced build, at 1.5 GW of contracted IT load with a multi-decade power purchase arrangement that ties into NEOM’s renewable and green-hydrogen architecture. The program’s structural distinctiveness is that it was scoped as renewables-first from day one rather than retrofitted onto a legacy gas-grid posture; the implication for AI-workload economics is that DataVolt’s energy mix and PUE profile will diverge from the rest of the Saudi fleet over time.

Hexagon, the data-center program with origins inside the Aramco perimeter, is reported at roughly 480 MW and is increasingly aligned to Humain’s offtake. Its differentiation is industrial-grade reliability and a power architecture that leverages Aramco’s gas backbone. Center3, the Saudi Telecom Company (STC) subsidiary, runs the connectivity-anchored fleet — colocation campuses in Riyadh, Jeddah, and the Eastern Province that do double duty as cable-landing and peering hubs.

The xAI training-cluster footprint, disclosed in 2025, gave Saudi Arabia a frontier-model anchor outside the Microsoft-OpenAI and Google-Anthropic axes. AWS’s Saudi region, anchored on $5.3 billion of disclosed capex, brings hyperscaler-grade availability-zone architecture and the accompanying control-plane investments. Microsoft Azure’s Saudi posture is being routed through both organic capacity build and the company’s $1.5 billion G42 stake, which now intersects with Humain-aligned capacity. Google Cloud’s commitment runs through both a Saudi region and direct AI-development collaboration with Humain. Oracle Cloud Infrastructure’s Riyadh region is operational. IBM’s footprint is smaller and more enterprise-focused.

Power Architecture

The most important number for any Saudi data-center campus is not its IT-MW capacity, but the power architecture beneath it. Saudi Electricity Company (SEC) is the regulated transmission and distribution monopoly; ACWA Power is the dominant independent power producer. Aramco supplies gas to most peaking and combined-cycle generation. Solar generation, anchored on the Sudair, Shuaibah, and Al Shuaiba PV programs, is the fastest-growing block on the supply curve. NEOM’s Helios green-hydrogen project, with 4 GW of associated renewables, sits as long-duration optionality in the regional energy mix.

Saudi data centers are being engineered against a peak ambient temperature regime that routinely exceeds 50C. That constraint propagates upward into thermal architecture (liquid cooling has moved from premium feature to baseline assumption for AI-grade campuses), water economics (closed-loop and air-cooled-condenser configurations dominate over evaporative-tower designs that are common in temperate climates), and power architecture (genset and UPS sizing has to assume worst-case ambient operation). The PUE numbers reported at Saudi facilities therefore have to be read with the climate context in mind; a 1.25 PUE in Saudi Arabia is a different engineering achievement than the same number in Iceland.

Sovereignty Posture

Every Saudi data-center campus sits on a sovereignty posture that is more legible than at most non-Gulf hyperscaler regions. The Cloud Computing Special Economic Zone, administered through ECZA, provides a regulatory umbrella under which hyperscalers can operate Saudi-resident regions while maintaining global control planes. The Personal Data Protection Law (PDPL), enforced by SDAIA, governs personal-data residency and cross-border transfer. The CST regulates communications and connectivity. SDAIA itself sits as the AI policy and data authority and operates the National Data Bank for state-classified data.

The practical effect is that hyperscalers operating in Kingdom run a tighter perimeter than they would in most non-Saudi markets. Customer data classified by Saudi data classification standards stays inside the Cloud SEZ envelope. Government workloads can be routed through SDAIA-managed sovereign capacity. Hyperscaler control-plane access is engineered with explicit data-residency exceptions. These sovereignty controls are the mechanism by which Saudi Arabia extracts both the operational benefits of hyperscaler infrastructure and the policy benefits of Kingdom-resident control.

In-Kingdom vs Regional

The cleanest analytical line in the Saudi infrastructure picture is the in-Kingdom versus regional distinction. In-Kingdom capacity sits behind the Saudi sovereignty perimeter; it is governed by PDPL, sits inside the Cloud SEZ where applicable, and is subject to Kingdom-resident audit. Regional capacity — typically operated out of UAE, Bahrain, or onward through European hyperscaler regions — is subject to its own jurisdictional controls and is not by default Saudi-sovereign.

The strategic implication is that the hyperscaler competition for Saudi workloads is increasingly being decided by which provider has Kingdom-resident capacity at sufficient scale. AWS’s Saudi region, Microsoft Azure’s Kingdom posture, and Google Cloud’s announced commitment all collapse the in-Kingdom-versus-regional distinction in their favor; providers without Kingdom-resident regions face a structural disadvantage on sovereign-AI tenders.

Pipeline and What’s Real

Of the 6.6 GW of disclosed capacity, the operational subset at the time of writing is in the low single-digit gigawatts. The next eighteen-month window, broadly through end of 2027, is where the largest tranches of announced capacity convert from civil works into energized IT load. The leading indicators to watch on a per-campus basis are: the date of the first power-on test, the date of the first hyperscaler availability-zone certification, the date of the first GPU-cluster acceptance test, and the ratio of contracted-to-deployed accelerators inside the campus.

Saudi infrastructure is now far enough along that the announcement-versus-delivery question can be answered with measured numbers rather than glossy renderings. The platform tracks every campus on those leading indicators, with site-level detail on power, cooling, sovereignty, and accelerator footprint.

Cooling and Thermal Architecture

Liquid cooling is the operational baseline for AI-grade campuses inside the Kingdom. The reference architectures across the major operators have converged on direct-to-chip liquid loops paired with rear-door heat exchangers and rack-level coolant distribution units. Single-phase water-glycol mixes dominate the deployed footprint; two-phase immersion is in pilot at selected sites but has not yet reached scale deployment. The thermal-architecture choice is forced by the combination of 50C-plus peak ambients and the rack-level power densities that GB300 NVL72 reference designs assume — air cooling at those densities and ambients is not an engineering option in the regional climate.

The water-economics picture is the second-order effect of the thermal choice. Closed-loop water systems and air-cooled-condenser configurations dominate over evaporative-cooling architectures that are common in temperate-climate hyperscale builds. The water consumption per megawatt-year of IT load at Saudi-resident campuses is therefore meaningfully lower than at, say, a US-Pacific-Northwest hyperscale region; the trade-off is the higher capex on the chiller-plant and the somewhat higher electrical-overhead consumption that closed-loop architectures impose.

NEOM-resident capacity is the most distinct thermal architecture. The DataVolt program’s renewables-anchored design pairs the thermal architecture with the broader Helios green-hydrogen and renewables footprint, with district-cooling optionality at scale that the inland and Eastern Province campuses do not have. The thermal architecture is therefore a function of the campus’s broader energy-systems integration, not just the individual data-center engineering choice.

Connectivity and Cable Landings

Saudi-resident compute capacity sits on a connectivity backbone that has scaled materially over the program window. The major submarine cable systems landing on Saudi coasts include the SEA-ME-WE 5 and SEA-ME-WE 6 systems on the Red Sea, the FALCON cable on the Gulf, the Saudi Vision Cable that connects the Red Sea and Gulf coasts overland, and the broader regional cable architecture that ties Saudi Arabia to East Africa, the Mediterranean, South Asia, and the Indian Ocean basin. STC’s Center3 subsidiary anchors the cable-landing-station footprint and operates the carrier-neutral colocation capacity that pairs with the cable infrastructure.

The fiber-backbone density inside the Kingdom has expanded materially under MCIT’s Vision 2030 connectivity workstream. The cross-Kingdom backbone connects the four primary clusters at backbone-grade fiber capacity; the metropolitan fiber footprints in Riyadh, the Eastern Province, Jeddah, and the NEOM area are sufficient to support hyperscale availability-zone architectures inside the metropolitan envelope. The connectivity layer is therefore not a binding constraint on the capacity buildout in the operationally relevant time horizon.

The Operator Landscape

Beyond the marquee anchor operators, the Saudi infrastructure landscape includes a growing tier of mid-cap and specialized operators. The Tonomus subsidiary inside NEOM operates technology infrastructure at the giga-project level. The DataVolt program operates as a build-and-operate counterparty to NEOM and to Humain. Mobily’s data-center footprint, paired with the broader Mobily commercial enterprise, anchors a connectivity-led colocation business. Salam (Etihad Atheeb) operates fiber and connectivity at the carrier layer with adjacent colocation capacity. The broader colocation and managed-services ecosystem includes a long-tail of regional and specialized operators that round out the supply side.

The competitive dynamic across the operator landscape is increasingly anchored on hyperscaler tenancy. The colocation and wholesale capacity that hyperscalers absorb for their Kingdom-resident regions is the largest single demand block; the sovereign-AI capacity that Humain absorbs is the second; the enterprise-and-government colocation market is the third. The platform tracks the per-operator demand-and-supply picture as it evolves.

Sustainability and Reporting

Sustainability disclosure is increasingly part of Saudi-resident campus operations. Operators report Power Usage Effectiveness (PUE), Water Usage Effectiveness (WUE), and Carbon Usage Effectiveness (CUE) on a per-campus basis where the disclosure framework supports it; the broader Vision 2030 environmental targets and the Saudi Green Initiative establish the policy frame inside which the per-campus disclosures sit. Renewables-anchored campuses (notably DataVolt’s NEOM program) report meaningfully different sustainability profiles than gas-grid-connected campuses, and the differential is itself an analytical input into the per-campus comparative work.

The reporting cadence is settling into an annual rhythm aligned with the broader Saudi corporate disclosure calendar. The platform tracks per-campus sustainability disclosures as they surface and integrates them into the infrastructure analytical layer. The aggregate sustainability picture across the 6.6 GW disclosed pipeline is itself a leading indicator for whether the program meets the broader Vision 2030 environmental targets while delivering the AI-infrastructure capacity envelope.

Permitting and Civil Works

Permitting cadence at the campus level is the most consequential leading indicator for the announced-to-energized ratio. The Royal Commission for Riyadh City, the Royal Commission for Jubail and Yanbu, the NEOM regulatory framework, and the broader municipal-and-provincial permitting layers each handle a share of the campus-permitting workflow. The Cloud SEZ administrative framework streamlines a subset of the workflow for hyperscaler-resident builds, but does not eliminate the underlying civil-works permitting cycle.

Civil-works execution at the campus level is anchored on a small number of major contractor relationships. The general-contractor and electrical-contractor pools are dominated by Saudi-domiciled firms with deep Vision 2030 program-execution experience, paired with international engineering-and-construction partners that bring the AI-grade campus engineering specialization. The platform tracks civil-works status as it surfaces.

For deeper reading: