$100B Project Transcendence

Saudi Arabia launched the $100 billion Project Transcendence AI initiative. SDAIA and NVIDIA deploying 5,000 Blackwell GPUs for sovereign AI factory. $14.9B in AI investments announced at LEAP 2025. Hexagon Data Centre — world's largest government DC at 480 MW — broke ground January 2026. $10B AMD agreement for multi-exaflop computing centers.

Sources

Saudi Cabinet Declares 2026 Year of AI — Hexagon DC Breaks Ground

Saudi Cabinet formally designates 2026 the Year of Artificial Intelligence. The $2.7B Hexagon Data Center project — described as the world's largest government data center by megawatt capacity — breaks ground in Riyadh in January 2026. $9.1B flows into 664 companies in the same year. HUMAIN's multimodal Arabic LLM continues development as PIF's primary sovereign AI vehicle. Project Transcendence's 180,000 H100 GPU compute base (500 MW) remains the operational backbone.

Sources

HUMAIN + NVIDIA AI Factories — 500MW, GB300 Shipments Begin

HUMAIN (PIF subsidiary) announced strategic partnership with NVIDIA to build AI factories with up to 500MW capacity powered by several hundred thousand GPUs over five years. First phase: 18,000 NVIDIA GB300 Grace Blackwell supercomputer with InfiniBand networking. Riyadh and Dammam facilities targeting Q2 2026 launch at 100MW each. HUMAIN received first GPU shipments in early 2026. Separately, HUMAIN unveiled HUMAIN OS, an agentic AI-powered operating system for enterprise workflows running on sovereign gigawatt-scale compute. A $1.2B framework with Infra signed to hyperscale Saudi AI data center capacity.

Sources

HUMAIN-xAI 500MW AI Data Center + NVIDIA 600K GPU Expansion

HUMAIN and xAI announced a partnership to build a flagship 500MW+ hyperscale GPU data center in Saudi Arabia, with xAI deploying Grok models nationally. Separately, HUMAIN expanded its NVIDIA partnership to deploy up to 600,000 GPUs over three years. AWS will also deploy 150,000 NVIDIA GPUs in a dedicated Riyadh AI Zone. xAI will deploy approximately 18,000 NVIDIA GB300 GPUs in the first cluster.

Sources

NEOM pivots to data center hub; HUMAIN targets 6.6GW decade roadmap

NEOM megaproject is being scaled back and redesignated as a data center hub, with a $5B DataVolt partnership for water-cooled infrastructure. Meanwhile, HUMAIN has outlined a decade roadmap targeting 6.6GW of AI compute capacity. Qualcomm AI200/AI250 rack services targeting 200MW starting 2026. PIF's 2026-2030 strategy shifts emphasis from The Line to AI infrastructure.

Sources

HUMAIN Expands with xAI, AWS, Global AI, NVIDIA at US-Saudi Forum

HUMAIN expanded NVIDIA partnership to deploy up to 600,000 GPUs (GB300) across Saudi and US data centers. xAI building 500MW data center in Saudi (first outside US, ~18K GB300 GPUs). AWS deploying 150K NVIDIA GPUs in dedicated Riyadh AI Zone. Global AI partnership for large-scale US data centers. HUMAIN ONE launched as enterprise-grade agentic AI OS powered by AWS. Total investment commitments now exceed $20B.

Sources

HUMAIN-Qualcomm 200MW Inference Rack Deployment

HUMAIN targeting 200MW of Qualcomm AI inference racks from 2026, diversifying beyond NVIDIA for inference workloads. Qualcomm partnership announced Oct 2025 with deployment beginning 2026.

Sources

NEOM-DataVolt $5B AI Datacenter Campus in Oxagon — Red Sea Cooling

NEOM announced a $5B partnership with DataVolt to build an AI data center campus in NEOM's Oxagon district. The site will use Red Sea seawater for cooling, enabling net-zero operations without fresh water consumption. DataVolt is developing a 1.5GW 'AI factory' at Oxagon. Phase one is expected operational by 2028. This follows the cancellation of The Line after $50B spent, with NEOM pivoting to AI infrastructure.

Sources