Saudi Infrastructure

The Physical Foundation of Saudi AI

Every AI model that runs in Saudi Arabia — whether it trains on Arabic-language data in an SDAIA facility, serves inference requests for a Aramco engineering application, or processes government transactions in a Humain sovereign cloud — ultimately depends on physical infrastructure: data centers built from steel and concrete, power grids fed by solar arrays and gas turbines, fiber optic cables buried under desert and ocean floor, and cooling systems that fight a climate where ambient temperatures regularly exceed 45 degrees Celsius.

The Saudi infrastructure sector is where the $77 billion AI buildout becomes concrete. It is the sector that converts capital commitments into watts of power, square meters of raised floor, and terabits of network throughput. With 12 entities tracked and a total capex of $508.4 billion — the second largest total of any sector — Saudi infrastructure represents the largest single-asset investment category in the entire buildout program.

Understanding this sector requires understanding the particular advantages and challenges that Saudi Arabia’s geography, climate, and capital structure present for large-scale AI infrastructure development.

Saudi Arabia’s Infrastructure Advantages

Energy Abundance at Low Cost

Saudi Arabia’s most significant infrastructure advantage for AI compute is energy. AI training workloads are among the most energy-intensive computing tasks humans have ever attempted — a single NVIDIA GB300 NVL72 rack consumes approximately 120 kilowatts, and a 100,000 GPU cluster requires roughly 600 megawatts of dedicated power. At US commercial electricity rates of $0.07-0.12 per kilowatt-hour, the energy cost of operating such a cluster over three years exceeds $1 billion. In Saudi Arabia, the combination of cheap hydrocarbons, aggressive renewable deployment, and government-subsidized industrial electricity rates produces energy costs that are among the lowest in the world for large industrial consumers.

ACWA Power, which holds an SCS of 8.1 and is controlled by PIF, is the critical link between Saudi energy abundance and AI infrastructure demand. ACWA is one of the world’s largest independent power producers, with over 60 gigawatts of capacity in development globally. Its Saudi projects include some of the world’s lowest-cost solar generation — the Al Shuaibah PV plant achieved a record-breaking $0.0104 per kilowatt-hour levelized cost of energy at financial close. For AI data centers, which are essentially large electricity consumers, access to ACWA-developed power at rates approaching one cent per kilowatt-hour is transformative for the economics of Saudi compute.

The NEOM Oxagon project, where DataVolt is developing 1.5 gigawatts of net-zero AI compute capacity, illustrates the power-AI nexus in its most ambitious form. NEOM has committed to 100% renewable energy supply, drawing on the $5 billion NEOM Green Hydrogen project and dedicated solar capacity to power what would be one of the world’s largest net-zero AI factories. The combination of ultra-cheap renewable generation, zero land cost for solar arrays in the Tabuk desert, and a captive industrial buyer creates an energy cost profile that no European or American data center developer can replicate.

Land Availability and Low Seismic Risk

Saudi Arabia has essentially unlimited land available for large-footprint infrastructure development. The Rub’ al Khali (Empty Quarter) alone is larger than France, and the kingdom’s overall land area is more than double that of Texas. For AI data centers, which require large flat sites with clearance for cooling infrastructure, backup power systems, and future expansion, Saudi Arabia’s land availability translates into near-zero site acquisition cost and the ability to plan multi-gigawatt campuses without the constraints of urban land markets.

The kingdom’s seismic profile is also favorable. Unlike Japan, the US West Coast, or Turkey — all significant data center markets — Saudi Arabia sits on the stable Arabian Plate with minimal earthquake risk across most of its territory. This reduces both construction costs (seismic reinforcement requirements are minimal) and resilience requirements (backup systems can be sized for operational rather than catastrophic scenarios).

PIF Capital for Construction

The Public Investment Fund’s role in infrastructure construction financing is as important as its role in AI compute ownership. PIF has made infrastructure bankability its explicit strategy — by providing equity capital, sovereign guarantees, and anchor tenancy commitments for major projects, PIF reduces the financing risk for private infrastructure developers and enables projects that would not achieve financial close on pure commercial terms.

DataVolt’s Oxagon project, Center3’s 1.1 gigawatt colocation buildout, and the hyperscaler regional expansions all benefit from PIF’s capital ecosystem — whether through direct PIF equity, PIF-backed anchor customer commitments, or the sovereign guarantee that Saudi infrastructure projects will receive the regulatory approvals and utility connections they require. This capital structure is why Saudi infrastructure projects can move at a pace that infrastructure projects in less centralized markets cannot match.

The Data Center Buildout Trajectory

Center3: The Colocation Backbone

Center3 is a stc subsidiary and Saudi Arabia’s premier carrier-neutral colocation provider. Its SCS of 7.6 reflects a combination of strong execution capability (stc’s existing infrastructure network), solid capital backing (stc’s balance sheet and PIF indirect exposure through stc ownership), and strategic positioning as the neutral interconnection fabric for the Saudi cloud ecosystem.

Center3’s 1.1 gigawatt by 2030 development target is the most concrete large-scale colocation commitment in the Saudi market. Carrier-neutral colocation is the infrastructure layer that makes hyperscaler competition possible — when AWS, Azure, Google Cloud, and Oracle Cloud all need to build Saudi regions, they need interconnected facilities where their networks can peer with enterprise connectivity. Center3 provides that fabric, and its scale target of 1.1 GW suggests an expectation that the Saudi cloud market will be among the largest in the world by end of decade.

Center3’s most important facilities are concentrated in Riyadh, the kingdom’s capital and largest city, and in the Eastern Province’s Dammam-Khobar corridor, which hosts Aramco’s headquarters and Saudi Arabia’s largest concentration of industrial enterprise IT. A third emerging cluster is Jeddah, which serves as the gateway for subsea cable landings on Saudi Arabia’s Red Sea coast and is the primary connectivity hub for the western part of the kingdom.

DataVolt: The NEOM Net-Zero Factory

DataVolt’s 1.5 gigawatt development at NEOM Oxagon represents the most ambitious single-site AI compute project in the world. With a $5 billion investment commitment and a mandate to deliver net-zero AI compute at scale, DataVolt’s Oxagon facility is designed from the ground up as an AI-native data center campus rather than a retrofit of traditional enterprise colocation.

The Oxagon site, on NEOM’s industrial waterfront, benefits from Red Sea access for thermal management, dedicated renewable energy supply from NEOM’s solar and wind assets, and direct connectivity to NEOM’s proprietary fiber infrastructure. The net-zero commitment is operationally meaningful, not just marketing — the economics of Saudi renewable energy make net-zero data center operation genuinely cost-competitive with fossil-powered alternatives, unlike in most other markets where the green premium is significant.

DataVolt’s positioning in NEOM also reflects the broader NEOM strategy of using AI and digitization as the enabling infrastructure for a city that has no legacy systems to retrofit. Every municipal service, transportation system, and commercial operation in NEOM is being designed to run on AI-native platforms from the start. The 1.5 GW compute campus is not serving historical demand — it is pre-building the infrastructure for a digital city that does not yet exist.

Hexagon: The SDAIA Anchor

Hexagon’s 480 MW SDAIA-anchored development in Riyadh occupies a distinctive position in the infrastructure landscape. Unlike Center3’s commercial colocation model or DataVolt’s net-zero factory, Hexagon’s Saudi development is anchored by a sovereign customer commitment — SDAIA’s requirement for dedicated compute capacity to support Allam and the national AI program’s inference and training workloads.

The SDAIA anchor transforms the project’s risk profile. Where commercial data center developers must assemble enterprise customer commitments to achieve financing, Hexagon’s SDAIA relationship provides the credit quality and volume commitment that makes a 480 MW facility financially viable from day one. This structure reflects a broader pattern in Saudi infrastructure where sovereign customer commitments are being used to de-risk private investment — a hybrid public-private model that accelerates private capital deployment while ensuring sovereign requirements are met.

The Power-AI Nexus: ACWA Power’s Critical Role

Why Energy Is the Binding Constraint

The AI compute buildout’s ultimate binding constraint is not silicon, capital, or talent — it is power. A 1 gigawatt data center campus requires more electricity than some small countries consume for all purposes. Delivering that power requires not just generation capacity but transmission infrastructure, grid stability management, and backup power systems capable of maintaining compute uptime at 99.99% availability.

ACWA Power is the entity that makes gigawatt-scale AI infrastructure power supply feasible in Saudi Arabia. As one of the world’s largest renewable energy developers, ACWA brings both the project development expertise and the balance sheet credibility to build dedicated power infrastructure for AI compute campuses at the scale the Saudi buildout requires.

ACWA’s role is not to become an AI company. It is to ensure that the AI companies building in Saudi Arabia — Humain, DataVolt, Center3, the hyperscalers — have reliable, low-cost power supply that meets their sustainability commitments. ACWA’s pipeline of Saudi solar and wind projects, combined with its established relationships with Saudi Aramco for gas backup capacity, positions it as the energy supply chain anchor for the AI buildout.

The Renewable Credibility Factor

Saudi Arabia’s renewable energy ambitions — 50% renewable power by 2030 under Vision 2030 — are relevant to AI infrastructure in ways that go beyond ESG optics. Major hyperscalers including Microsoft, Google, and Amazon have made corporate commitments to 100% renewable energy that affect their data center procurement requirements. Building in a market where cheap, credible renewable energy is available from a development partner like ACWA Power allows hyperscalers to meet their energy commitments without the premium that renewable power commands in markets with less solar resource.

Saudi Arabia receives among the highest solar irradiance of any large country in the world — annual direct normal irradiance of 2,200-2,600 kilowatt-hours per square meter across most of the kingdom. Combined with the availability of large flat land for solar array installation, this makes Saudi Arabia a structurally low-cost renewable energy market that will only become more attractive as solar panel costs continue to decline.

Infrastructure Gaps and Challenges

Cooling in Extreme Heat

The most significant technical challenge for Saudi AI data centers is cooling. AI compute generates enormous heat — a GB300 NVL72 rack operating at 120 kW must have that heat removed continuously to prevent thermal throttling or hardware damage. In a climate where outdoor temperatures exceed 45°C for months at a time, air-side economization cooling (which works by drawing cool outside air through the facility) is not an option for most of the year. Saudi AI data centers must use mechanical cooling — chillers, cooling towers, or direct liquid cooling — for the majority of annual operating hours.

This requirement increases capital expenditure and power usage effectiveness (PUE) compared to data centers in cooler climates. A facility in Iceland or Oregon can achieve PUE of 1.05-1.10 through air-side cooling. A Saudi facility using chilled water or direct liquid cooling will target PUE of 1.25-1.35 under current best practices. The gap is narrowing as direct-to-chip liquid cooling technology matures — NVIDIA’s GB300 NVL72 rack is designed for rear-door liquid cooling that significantly reduces facility cooling requirements — but thermal management remains a meaningful cost factor.

DataVolt’s Oxagon facility’s location on the Red Sea coast is partly a response to this challenge. Access to seawater for heat rejection reduces the cooling infrastructure cost and enables lower PUE than fully landlocked sites.

Connectivity Between Regions

Saudi Arabia’s three major population and economic centers — Riyadh in the center, Jeddah on the Red Sea coast, and the Eastern Province on the Gulf coast — are separated by hundreds of kilometers of desert. Connecting AI compute facilities in these three clusters to each other and to international markets requires substantial fiber infrastructure investment.

Current connectivity between regions relies on Saudi Telecom Company’s national fiber network, supplemented by competing carrier networks. Subsea cable capacity is growing rapidly — Saudi Arabia’s Red Sea coast has become a major landing point for cables connecting to Europe and Africa, while Gulf coast landings provide connectivity to India and Asia. However, the concentration of AI compute in Riyadh, which is the most inland of Saudi Arabia’s major cities, requires long-haul fiber to reach international submarine cable systems, adding latency that global hyperscalers must manage.

Saudization Requirements

Saudi Arabia’s Nitaqat program requires that companies operating in the kingdom meet minimum Saudi national employment thresholds, with requirements varying by industry. For data center operations, Saudization requirements mean that international infrastructure companies cannot simply deploy expat operations teams — they must recruit, train, and retain Saudi national employees for technical roles, including data center technicians, power systems engineers, network operations staff, and security personnel.

This requirement is both a challenge and an opportunity. The challenge is that Saudi Arabia’s technical workforce is still developing the depth of AI infrastructure expertise that international data center operators rely on from established talent markets. The opportunity is that companies that invest in Saudi talent development — through partnerships with Saudi universities, apprenticeship programs, and structured technical training — build both regulatory compliance and lasting competitive advantage in the market.

What International Infrastructure Companies Need to Know

For EPC contractors, equipment suppliers, and operations firms seeking to participate in the Saudi infrastructure buildout, several practical realities shape the market.

The scale of projects is exceptional. Individual data center campuses are being planned at 480 MW, 1.1 GW, and 1.5 GW — larger than any single-site data center project ever completed in North America or Europe. Companies whose project delivery experience maxes out at 50 MW will need to scale their capabilities, their supply chains, and their project management systems to compete.

Procurement is relationship-intensive and moves at sovereign speed. Major projects are typically awarded through negotiated processes involving sovereign entities rather than purely competitive tender. Relationships with Humain, PIF, SDAIA, and ACWA Power are the essential entry points for large project opportunities. Companies without established Saudi relationships should pursue agent or JV structures with Saudi entities that have existing government access.

Local content requirements are increasing. Saudi Arabia’s Vision 2030 program explicitly targets the development of domestic manufacturing and services capability. Suppliers of data center equipment, power systems, cooling infrastructure, and construction services will face increasing expectations to source components locally, manufacture in-kingdom, or partner with Saudi companies. The Saudi government’s parallel effort to build domestic semiconductor and hardware manufacturing through ALAT is an early signal of the direction of local content policy.

The infrastructure buildout timeline is real and compressed. Saudi Arabia is not building AI infrastructure on a ten-year academic planning horizon — Humain’s Phase 1 GPU deployment is in-progress, Center3’s facilities are under construction, and ACWA Power’s dedicated AI campus power projects are in development. Companies that can mobilize quickly, with pre-qualified supply chains and experienced project teams, will capture the near-term opportunity. Companies that require multi-year qualification processes will miss the first wave.

The Saudi infrastructure sector is not merely a construction program. It is the physical expression of a sovereign decision to build AI capability at national scale. The companies that understand the strategic logic behind the infrastructure investment — not just the procurement specifications — will be the ones that build enduring positions in one of the world’s most important infrastructure markets.