Decoding the Sovereign Compute Score: Who Actually Holds Saudi AI Power
The Sovereign Compute Score (SCS) is a composite metric designed to measure what matters in a sovereign AI buildout that is structurally different from a commercial hyperscaler expansion. It is not enough to have capital. It is not enough to have government mandate. SCS attempts to capture whether an entity can actually convert ambition into operating AI infrastructure under sovereign control.
The seven components — Capacity (18%), Capital (16%), Silicon Access (16%), Sovereignty (13%), Geopolitical Resilience (13%), Velocity (12%), and Execution (12%) — are weighted to reflect the specific constraints of AI infrastructure development in a Tier-2 jurisdiction under US export controls, where capital and sovereignty must coexist with hardware acquisition complexity.
Why the Weights Are What They Are
Capacity (18%) leads because megawatts of deployed AI compute is ultimately the scorecard. A sovereign AI program exists to run AI workloads at scale; capacity is the direct measure of whether that is happening.
Capital (16%) and Silicon Access (16%) are co-equal second tier. Capital without silicon is a sovereign wealth fund with empty buildings. Silicon without capital is an engineering team without a facility. The joint constraint is the binding one.
Sovereignty (13%) and Geopolitical Resilience (13%) measure whether the AI capability is genuinely independent — whether a foreign government’s export control decision, sanctions action, or vendor withdrawal could disable the program. This is what distinguishes a sovereign compute program from commercial cloud usage.
Velocity (12%) and Execution (12%) capture the speed and demonstrated track record of delivery. A perfect score on the first five components means nothing if the buildout is perpetually delayed.
Humain 9.3: The Full-Stack Sovereign AI Operator
Humain’s 9.3 SCS is the highest not because it is the largest entity in Saudi AI but because it most completely satisfies all seven components simultaneously. Launched in May 2025 with PIF backing, Humain is purpose-built to be a sovereign AI operator: its mandate is to own and operate AI infrastructure, deploy AI platforms, and develop AI applications — all under Saudi sovereign control.
On capacity, Humain has the largest MW commitment (1,900 MW). On capital, PIF’s backing provides virtually unlimited patient capital. On silicon access, the 18,000 GB300 GPU Phase 1 and the AMD-Cisco-Humain JV for 1 GW over 5 years give Humain multi-vendor silicon diversification. On sovereignty, as a PIF subsidiary Humain sits entirely within Saudi state ownership with no foreign equity. On geopolitical resilience, the diversified silicon strategy (NVIDIA + AMD + Qualcomm inference + Groq) reduces single-vendor dependency. On velocity, the Humain launch was followed within weeks by announced deals totaling tens of billions in commitments. On execution, while Humain is new, Tareq Amin’s background building Rakuten Mobile and DISH Network’s cloud-native infrastructure gives the organization operational credibility.
The 0.7 gap from a perfect 10.0 reflects that Humain is new — execution track record at scale has not yet been established, and some announced capacity is still in planning rather than build phase.
ALAT 8.6: Capital Anchor and Manufacturing Ambition
ALAT (Advanced Technology company) scores 8.6 — second overall — for a reason that is distinct from Humain. Where Humain is an AI operator, ALAT is a hardware manufacturing and deep tech investment vehicle owned by PIF. ALAT’s strategic relevance to sovereign AI is that it is attempting to build indigenous semiconductor and advanced technology manufacturing capability in Saudi Arabia.
ALAT’s capital scores near-maximum: PIF funding means access to tens of billions. Its sovereignty score is near-perfect: 100% Saudi-owned. Its geopolitical resilience benefit is explicit — if Saudi Arabia can manufacture some fraction of its own AI silicon domestically, its vulnerability to US export controls diminishes structurally.
ALAT’s lower scores on velocity and execution relative to Humain reflect that manufacturing semiconductor capability is a decade-long program, not a three-year buildout. The ambition is credible (ALAT has announced investments in chip design, packaging, and advanced manufacturing), but the gap between investment announcement and operating fab is very large.
For investors and vendors, ALAT represents the longest-horizon bet in Saudi AI. If ALAT successfully develops Saudi semiconductor capability by 2035, the geopolitical implications for global chip supply chains are significant. If it stalls, ALAT remains a capital vehicle rather than a manufacturing reality.
PIF 8.5: The Capital Anchor of the Entire Ecosystem
The Public Investment Fund scores 8.5 as an entity distinct from its portfolio companies. This is the right analytical framing: PIF is not simply Humain’s parent — it is the funding source, strategic coordinator, and ultimate sovereign backstop for essentially the entire Saudi AI buildout.
PIF’s capital score is perfect. Its sovereignty score is near-perfect (it operates as Saudi Arabia’s primary sovereign wealth vehicle under direct royal oversight). Its geopolitical resilience score reflects that PIF’s financial flexibility allows it to pivot capital allocation faster than any other actor in the ecosystem.
PIF’s lower scores on capacity, silicon access, velocity, and execution reflect its nature as an investor and capital allocator rather than an operator. PIF does not run data centers; it funds entities that do. The SCS correctly discounts this one step removed from operational reality.
MCIT 8.1 and NCDAI 8.1: Governance as Infrastructure
The fact that two governance entities — the Ministry of Communications and Information Technology and the National Center for Digital AI — score 8.1, matching enterprise operators like Red Sea Global and Saudi Aramco, reveals something important about Saudi AI power structure: regulatory and policy control is as strategically valuable as operational compute.
MCIT under Minister Abdullah Al-Swaha has been the architect of the Cloud First Policy, the Special Economic Zone framework for tech companies, and the regulatory environment that determines which foreign providers can participate in Saudi cloud markets. MCIT scores high on sovereignty (government ministry with legislative authority), capital (government budget plus ability to direct sovereign investment), and velocity (Al-Swaha’s ministry has moved faster on tech policy than almost any comparable ministry globally).
NCDAI’s 8.1 reflects its role as the coordination body for Saudi digital economy and AI strategy — the entity that translates Vision 2030 AI ambitions into specific program mandates, budget allocations, and execution tracking. High on sovereignty, governance authority, and strategic alignment; constrained on direct capacity and silicon access scores.
The implication for foreign vendors: MCIT and NCDAI approval and relationship is not optional. These entities determine market access, certification requirements, and preferred vendor status. An AI infrastructure company that has Humain as a customer but has not built relationships with MCIT and NCDAI is strategically exposed.
SDAIA 6.6: The Surprising Mid-Range Position
SDAIA scores 6.6 — meaningfully lower than its prominence in Saudi AI discourse might suggest. This score deserves careful explanation because SDAIA is frequently described as the “AI authority” of Saudi Arabia and the home of Allam, the national Arabic LLM.
SDAIA’s mid-range position reflects a structural reality: it is a regulatory and operational body with significant authority in data governance and national AI strategy, but it is not the capital anchor (PIF/Humain is) and its operational compute is limited (480 MW target, 5,000 Blackwell GPUs in current deployment). SDAIA’s silicon access is government-procurement limited rather than commercially flexible. Its velocity score is moderate — SDAIA moves at government pace, not commercial pace.
The 6.6 score is not a criticism of SDAIA’s importance. It is an accurate reflection that in the specific dimensions that determine sovereign compute power — hardware procurement at scale, capital deployment speed, technical execution velocity — SDAIA sits in the middle tier. Its authority in data policy and Arabic AI development is high; its direct contribution to the MW buildout is smaller than the headline numbers suggest.
stc 7.9 and Aramco 7.9: Infrastructure Anchors
Saudi Telecom Company and Saudi Aramco both scoring 7.9 reflects their position as infrastructure anchors rather than frontier AI operators. stc’s score is driven by its ownership of the national telecom backbone, Center3 data center operations, deep government relationships, and the Humain JV equity. stc is a 7.9 because it is indispensable infrastructure — but its AI-native ambition is partially outsourced to the Humain JV rather than developed wholly organically.
Aramco’s 7.9 reflects the world’s most data-rich industrial company applying AI to the world’s most consequential industry vertical. The Groq-Aramco Digital partnership (world’s largest inference facility outside the US, $1.5 billion commitment) is a direct sovereign compute investment. Aramco’s score is constrained by the fact that its AI program, while massive in absolute terms, is primarily industrial/operational AI rather than sovereign AI infrastructure for third-party use.
Groq 6.4: Specialist Infrastructure Play
Groq’s 6.4 is the only foreign company in the top 10, and its position requires explanation. Groq’s LPU (Language Processing Unit) inference silicon has achieved extraordinary token generation rates — approximately 500,000 tokens per second at the Aramco Digital facility — making it the world’s fastest inference infrastructure at commercial scale.
Groq scores 6.4 rather than higher because: it is a US company with no Saudi ownership (sovereignty constraint), its geopolitical resilience is limited by US export control exposure, and its capital position is commercial rather than sovereign. The $1.5B Aramco Digital facility is a Saudi-anchored deployment, but the underlying silicon and platform IP remains in foreign hands.
For investors, Groq’s 6.4 score in a Saudi-dominated index is actually a strong signal: a US company without sovereign backing achieving this level of strategic relevance means its technology is genuinely differentiated, not just well-marketed.
What the SCS Distribution Reveals
The most important insight from the SCS distribution is the separation between capital-sovereign entities (Humain, ALAT, PIF) scoring 8.5-9.3 and operational-industrial entities (Aramco, stc, Center3) scoring 7.6-7.9. This gap is not about competence — it is about mandate.
Humain, ALAT, and PIF exist specifically to build Saudi sovereign AI capability. Aramco, stc, and Center3 are building AI capability for their core business verticals. The former group has sovereign AI as its primary objective function; the latter group has it as an important but not singular objective.
For foreign companies deciding where to invest capital and management attention in Saudi AI, the SCS distribution points to Humain and PIF portfolio entities as the primary partnership and vendor targets — they hold the mandate and the capital. Governance entities (MCIT, NCDAI, SDAIA) determine market access. Industrial operators (Aramco, stc) represent the enterprise AI revenue opportunity.
SCS as a Vendor Targeting Tool: Practical Applications
For AI vendors and infrastructure companies entering the Saudi market, the SCS distribution provides a practical targeting framework. The score tiers correspond to different vendor engagement strategies:
9.0+ (Humain): Full account team deployment justified. These entities have both the capital authority and the sovereign mandate to sign large, multi-year contracts. The sales cycle is long and relationship-intensive but the deal sizes are proportionally large. Vendor investment in Saudi-based teams, Arabic-language product adaptation, and local partnership structures are required and will be rewarded.
8.0-9.0 (ALAT, PIF, MCIT, NCDAI, Red Sea Global, Saudi Aramco, stc, SNB): Priority enterprise sales targets. Each has meaningful AI budgets and clear use cases. The vendor strategy differs: ALAT and PIF are investment/partnership discussions rather than product sales; MCIT and NCDAI are regulatory and ecosystem relationships; the enterprise operators (Aramco, stc, SNB) are traditional enterprise AI sales with long cycles and high barriers to displacement once established.
6.0-8.0 (SDAIA, Groq, Center3): Addressable market with specific positioning requirements. SDAIA is a government partner that requires a sovereignty-aligned pitch. Groq is a potential technology partner or competitor depending on the vendor’s product category. Center3 is a co-location and infrastructure customer for hardware and networking vendors.
Why SCS Scores Will Change Over 2025-2030
The SCS ranking is a snapshot, not a permanent ordering. Several shifts are predictable:
Humain’s score will increase as execution matures — its current 9.3 has a small execution penalty for being a new entity; as it deploys 18,000 GPUs and proves operational capability at scale, the execution component score will rise.
SDAIA’s score may increase if Allam achieves large-scale deployment and if the SDAIA Hexagon data center reaches full 480 MW operational capacity. The current 6.6 reflects gaps in capacity and velocity that operational execution would close.
New entities will appear: ALAT’s hardware manufacturing programs, if they produce operating semiconductor or advanced electronics manufacturing facilities, would give a new entrant to the 8.5+ tier that combines capital, sovereignty, and silicon production simultaneously — a combination no current entity fully achieves.
The SCS framework is most valuable not as a static ranking but as a dynamic tracker of how the Saudi AI power structure evolves as capital commitments convert to operational reality over the 2025-2030 period.