When you’d compare alternatives to Center3
Center3 operates as Saudi Arabia’s primary carrier-neutral colocation provider, occupying a narrow but critical segment of the Kingdom’s digital infrastructure stack. As a neutral colocation operator, Center3 offers Tier III and Tier IV certified data center facilities where multiple network operators, cloud providers, and enterprise customers can interconnect without being captive to a single telco or hyperscaler. Its subsea cable landing station assets give it a position in the international connectivity layer that most domestic Saudi infrastructure operators lack. For analysts mapping Saudi Arabia’s AI compute buildout, Center3 surfaces in comparisons because it represents the colocation and interconnection layer that hyperscale AI workloads depend on—the physical meeting point where Saudi Telecom Company’s national backbone, international subsea cables, and enterprise compute environments converge.
The corporate profile frames the comparison. Center3 was founded in 2014, is headquartered in Riyadh, and operates certified facilities in both Riyadh and Jeddah — the Kingdom’s primary business hub and its primary Red Sea port and cable landing geography respectively. The Riyadh facility anchors the Saudi Internet Exchange (SAIX), the primary peering point for internet traffic entering and leaving the Kingdom, which makes Center3 an operator of critical national internet infrastructure rather than a landlord for server hardware. Its operational campus runs at roughly 100 MW today, with a stated ambition of 1 GW of additional capacity by 2030 — a target that would transform Center3 into one of the Kingdom’s largest colocation platforms. One further fact matters for reading this comparison set: stc is Center3’s corporate parent, so the stc-versus-Center3 question is less about rival ownership than about which operating model — integrated telco cloud or carrier-neutral colocation — better fits a given workload.
The comparison queries for Center3 arrive from several directions. Data center investors evaluating Saudi Arabia’s infrastructure market want to understand how Center3’s carrier-neutral model compares to the integrated utility and telco players that dominate the Saudi digital infrastructure landscape. Technology companies assessing their Saudi market entry options need to know whether Center3’s Tier IV facilities are sufficient for AI workloads, or whether the power infrastructure, cooling density, and fiber redundancy required for GPU clusters demand a partnership with a utility-scale energy provider like ACWA Power or a national telco like stc. Hyperscalers like Microsoft, Google, and AWS—all of which have announced or are building Saudi cloud regions—are evaluating whether Center3’s colocation fabric is robust enough to serve as a neutral interconnection hub or whether Saudi Electricity Company’s infrastructure makes direct utility partnerships more reliable.
Center3 scores 7.6 on the SCS, reflecting solid but not dominant positioning across the seven dimensions. Its highest normalized scores are in Sovereignty—as a Saudi-owned operator, it is structurally aligned with the Kingdom’s data localization mandates—and in Execution, where its Tier III/IV certifications and cable landing station operations represent delivered infrastructure rather than announced commitments. Its constraints are in Capital and Silicon Access: as a colocation operator rather than an energy utility or a telco, Center3’s balance sheet does not support the multi-billion-dollar capex cycles required to build new hyperscale campus capacity on its own, and it does not directly procure AI silicon.
How to read the alternative rankings
The Saudi Compute Score’s seven dimensions map directly onto the operational requirements of AI compute infrastructure. For a colocation operator like Center3, the most relevant dimensions are Capacity (18%), which measures physical data center footprint and committed expansion plans; Sovereignty (13%), which reflects alignment with Saudi data localization requirements; and Execution (12%), which captures the track record of delivering certified facilities on time.
The Capacity weight is the largest because AI compute at scale is ultimately a physical-world problem. A 100-megawatt GPU cluster requires not just floor space and fiber, but power infrastructure capable of delivering that load reliably, cooling systems designed for the thermal density of modern AI accelerators, and physical security meeting the standards of sovereign AI workloads. Center3’s Tier IV facilities meet security and redundancy requirements, but the megawatt scale of hyperscale AI deployments is approaching the upper limits of what a pure colocation model can accommodate without utility-scale energy partnerships.
Capital at 16% captures a structural gap for Center3. ACWA Power, scoring 8.1, has a capital structure built for multi-billion-dollar energy infrastructure projects. stc, scoring 7.9, has the telco cash flows and balance sheet to fund national-scale network infrastructure. Center3’s capital profile is more constrained, making it dependent on JV structures or anchor tenant pre-commitments to fund large capacity expansions. Silicon Access at 16% is not directly relevant to a colocation provider’s own operations, but it is highly relevant to Center3’s customers—and a data center operator whose customers cannot get GPU allocations sees its occupancy growth constrained regardless of its own facility quality.
The remaining dimensions complete the profile. Geopolitical Resilience (13%) is a quiet strength: a Saudi-owned interconnection operator carries none of the export control exposure that silicon suppliers manage and none of the jurisdictional complexity that foreign hyperscalers bring; its principal resilience question is concentration, since a large share of the Kingdom’s peering fabric converges on a small number of Center3 facilities, making them systemically important. Velocity (12%) is respectable: the operational campus at roughly 100 MW represents delivered capacity rather than announced intent, though the 1 GW expansion ambition will test whether Center3 can sustain delivery pace at ten times its current footprint. Execution (12%) is the dimension where Center3’s certifications, its SAIX operations, and its cable landing station access all register as completed infrastructure — the strongest form of evidence the SCS framework recognizes.
When the alternatives become preferable
When power infrastructure is the binding constraint. AI GPU clusters are extraordinarily power-hungry. The H100 SXM5 draws 700 watts; a 10,000-GPU cluster needs 7+ megawatts of IT load plus cooling overhead. ACWA Power, as Saudi Arabia’s largest independent power producer with 43+ GW of installed capacity and direct relationships with the Saudi Electricity Grid, can deliver power infrastructure at scales that a colocation operator cannot. When a hyperscale AI data center project requires guaranteed power at 50-500 MW scale, ACWA Power’s ability to develop dedicated power generation and deliver it under long-term PPA structures makes it the stronger partner than Center3, which is a tenant on the grid rather than a grid-scale power developer.
When national network reach is required. stc’s national fiber backbone connects every major city in Saudi Arabia and extends into international markets through its own subsea cable investments and roaming agreements. Center3 has cable landing station assets, but stc’s network reach is broader, its customer relationships deeper, and its government mandate for national connectivity more direct. For AI compute deployments that require low-latency connectivity to endpoints across the Kingdom—enterprise branches, smart city sensors, Vision 2030 project sites—stc’s network fabric is more comprehensive than Center3’s interconnection hub model.
When grid reliability and frequency regulation matter. Saudi Electricity Company operates the Saudi national grid and is the ultimate guarantor of power quality for any facility plugged into it. For AI data centers where grid instability could corrupt training runs or interrupt inference serving, understanding SEC’s grid investment roadmap is more important than any individual colocation provider’s SLA. SEC’s planned grid modernization investments, funded in part through Vision 2030 allocations, directly determine whether ACWA Power’s IPP projects and Center3’s facilities can reliably support hyperscale AI workloads.
When the project requires a government-mandated anchor. Saudi government AI projects—including those funded through the National AI Strategy and the HUMAIN compute program—will be directed to facilities and infrastructure partners that meet sovereign data requirements and are embedded in Vision 2030’s procurement framework. stc and SEC have direct government relationships that give them preferred status for those mandated projects. Center3’s carrier-neutral model is valuable for commercial cloud providers, but sovereign government AI workloads may be directed toward state-adjacent telco and utility structures first.
When subsea cable ownership rather than landing station access is required. Center3’s cable landing station access gives it connectivity value, but owning the cable—as stc does through its international investment in submarine cable systems—provides more control over capacity, pricing, and route redundancy. For AI inference applications that depend on sub-100ms latency to European or Asian endpoints, cable ownership creates a quality-of-service differentiation that a cable landing station tenant cannot match.
The competitive tier breakdown
ACWA Power (SCS 8.1) leads the alternatives for a fundamental reason: AI compute is primarily an energy problem, and ACWA Power is Saudi Arabia’s most capable energy infrastructure developer. ACWA Power has developed, financed, and operates power plants across 13 countries, with a Saudi portfolio that includes major solar, wind, and thermal generation projects tied to the Vision 2030 energy transition. Its relevance to AI compute is growing rapidly as data center developers recognize that the limiting factor for large GPU clusters is not silicon or fiber but reliable, competitively priced electricity. ACWA Power’s ability to develop behind-the-meter renewable power generation for dedicated data center campuses—a model already emerging in the US and Europe—gives it a competitive position in Saudi AI infrastructure that Center3, as a power consumer rather than a power developer, cannot replicate. ACWA Power’s Capital score of 8.1 reflects its project finance expertise and its proven ability to raise large-scale infrastructure debt on international markets, not just its own balance sheet. For hyperscalers building dedicated Saudi AI campuses, ACWA Power is increasingly a required partner, not just a utility vendor.
stc (Saudi Telecom Company, SCS 7.9) combines national network reach, data center operations through its stc Cloud division, and government mandate in a way that positions it as both a complement to and a competitor of Center3. stc’s data center portfolio is growing rapidly, driven by demand from Vision 2030 projects and from hyperscalers building Saudi cloud regions. Unlike Center3’s carrier-neutral model, stc’s data centers sit on its own network backbone, which gives stc customers integrated connectivity advantages but creates carrier dependency for tenants who need multi-network access. stc’s higher SCS relative to Center3 reflects its broader capital base, its national network coverage, and its direct procurement relationships with technology vendors—advantages that compound in the current cycle where the Saudi government is directing digital infrastructure investment through Vision 2030-aligned channels where stc has preferred status. For enterprise customers who need both connectivity and compute in a single-vendor Saudi structure, stc’s integrated offering is often more practical than Center3’s carrier-neutral model, even if the latter provides more theoretical flexibility. The parent-subsidiary relationship between stc and Center3 adds a final layer: buyers are not choosing between hostile competitors but between two operating models under a common group umbrella, which means the practical decision usually reduces to whether carrier neutrality or integrated network bundling better serves the workload.
Saudi Electricity Company (SCS 7.9) is the infrastructure substrate on which both Center3 and ACWA Power depend. SEC’s relevance to AI compute analysis lies primarily in its grid investment roadmap—specifically, the planned upgrades to transmission capacity, frequency regulation, and smart grid capabilities that will determine whether Saudi data centers can reliably support the power loads required for large-scale AI training. SEC’s SCS of 7.9 reflects strong Sovereignty and Capital scores: it is the government-controlled utility with the mandate and the funding to modernize Saudi grid infrastructure, and its investment program is backed by Vision 2030 capital allocations that insulate it from normal commercial risk constraints. For AI infrastructure planners who need to understand whether Center3’s facilities can handle 50+ MW AI campuses five years from now, SEC’s grid upgrade trajectory is the critical variable—and SEC itself, as the grid operator, has more direct influence over that trajectory than Center3.
The rest of the field
NEOM (SCS 7.9) enters the comparison not as a colocation provider but as the Kingdom’s most ambitious hosting environment for AI capacity. The $500 billion megacity’s most credible compute component is the DataVolt AI factory at Oxagon — a $5 billion, 1.5 GW net-zero data center development targeting 2028, powered by dedicated renewable generation. That single project, if delivered, would exceed the aggregate capacity of the Kingdom’s carrier-neutral colocation stock by an order of magnitude. For buyers, NEOM is not a substitute for Center3’s interconnection services; it is an illustration of where hyperscale AI capacity is being built outside the colocation model entirely.
Oxagon (SCS 7.4), NEOM’s industrial city on the Red Sea, is the one entity in this field that could eventually replicate Center3’s combination of cable landing access and colocation. As a subsea cable landing point and the designated AI factory zone within NEOM, Oxagon’s long-term positioning includes exactly the international connectivity layer that Center3 monetizes today. The difference is time: Oxagon’s infrastructure is largely planned while Center3’s is operational.
Mobily (SCS 7.2) is the most direct functional alternative on the connectivity layer. Saudi Arabia’s second carrier — with roughly 30% mobile market share and e& (formerly Etisalat) as a significant shareholder — operates its own cable landing infrastructure in Jeddah with live traffic across multiple subsea systems. For buyers whose priority is international backhaul diversity rather than neutral interconnection, Mobily’s Saudi-listed landing assets are a genuine substitute for Center3’s connectivity services.
Zain KSA (SCS 7.2), the third carrier at roughly 25–27% mobile share and part of Kuwait-based Zain Group’s nine-country footprint, competes for enterprise connectivity-plus-cloud bundles rather than for the interconnection hub role. Its relevance here is as a pricing and redundancy check: a three-carrier market keeps connectivity economics competitive for every colocation tenant.
Hexagon (SCS 6.4) is the sovereign pole of the comparison. At 480 MW, the SDAIA-operated facility is the world’s largest government data center, hosting the National Data Lake’s 430+ integrated government systems and a sovereign AI factory built on 5,000 NVIDIA Blackwell GPUs. It is not commercially available and never will be — but it defines the standard for sovereign workloads that Center3’s commercial facilities are not designed to serve, and it absorbs precisely the government demand that will never appear in the commercial colocation market.
Center3’s structural position
Center3 occupies a valuable but narrow structural position in Saudi Arabia’s AI compute ecosystem. Its carrier-neutral colocation model is genuinely important: in any national digital infrastructure ecosystem, there needs to be neutral ground where competing networks can interconnect and where cloud providers can colocate without being forced into exclusive relationships with a single telco. Center3 fills that role in Saudi Arabia and its Tier III/IV certifications make it suitable for enterprise and cloud workloads.
The structural challenge for Center3 in the AI compute cycle is that the workloads driving data center investment are not traditional enterprise workloads—they are GPU-intensive AI training and inference clusters that require power densities, cooling configurations, and capital commitments that strain the colocation business model. Hyperscalers building their own campuses, utility-scale power developers like ACWA Power creating integrated energy-plus-compute solutions, and national telcos like stc building integrated connectivity-plus-compute offerings are all potentially displacing the pure-play colocation model at the high end of the market. Center3’s strongest future position is as the interconnection hub that ties all of these players together—the neutral exchange point where hyperscaler, telco, and enterprise networks meet—rather than as a direct competitor to utility-scale data center developers.
The near-term reality is more favorable than the strategic challenge suggests. Center3’s roughly 100 MW of operational capacity is deliverable today, its SAIX operations make it indispensable to Saudi internet routing regardless of how the hyperscale layer evolves, and its 1 GW expansion target gives it a credible path to participating in the AI cycle rather than being bypassed by it. Whether that expansion is delivered — and delivered with the utility partnerships that ACWA Power and Saudi Electricity Company make possible — will determine whether Center3’s next SCS revision moves up or down.