Saudi Arabia's AI pitch: Build your data centres here. We'll make the electricity cheap.
US energy bills are climbing. Data centres are a big reason. Saudi Arabia sees the pain and is making a simple offer: bring your AI infrastructure to the kingdom, and your power costs drop.
"Saudi Arabia has one real comparative advantage for AI: cheap electricity," said Greg Priddy of the Center for the National Interest. By most other metrics, you might build elsewhere. But energy cost is the choke point, and that's where Riyadh leans in.
The scale Saudi is aiming for
Datavolt is building a $5bn data centre on the Red Sea coast. Humain, backed by the $1 trillion Public Investment Fund (PIF), plans a network from Riyadh to Dammam targeting 6.6 GW by 2034.
To put that in context, Pierre Pinson of Imperial College London notes that level of demand is roughly one and a half times Denmark's annual electricity use. Saudi's grid must expand to feed it-but "in five years, this is not mission impossible," he said.
Why this can work: the energy math
Data centres are electricity machines with cooling problems. Companies will use any reliable electrons they can source at scale. In the US, that increasingly means on-site gas turbines.
Saudi Arabia generates about 60% of its electricity from natural gas. More than a third still comes from burning crude and fuel oil. Solar is around 2% and rising fast.
Prices that tilt the ROI
Saudi gas is cheap. Electricity is cheaper than Europe and often even the US. Roland Berger estimates commercial power prices are 30-50% below the global average-an immediate lift to data centre margins.
Aramco's move to take a significant minority stake in Humain shows how energy and AI are being stitched together. As Aramco CEO Amin Nasser put it: "Data centres need lower-cost energy… this is the place for it."
The oil wildcard
Ellen Wald, author of Saudi Inc., makes the blunt point: if Saudi burns crude to run data centres, the marginal cost is near zero for the state. Production costs sit under $10 per barrel.
With global supply loosening, the opportunity cost of burning a slice domestically can be low. The International Energy Agency has flagged a sustained surplus in recent reports, which only strengthens the case for local use over exports in specific scenarios. IEA Oil Market Report
Policy lever: price control
Saudi Arabia can set electricity prices by decree. That's not how the US or Europe work. In liberalized markets, data centre clusters push up local prices; the grid passes constraints straight to consumers.
In the kingdom, the state can shield both operators and households. Pinson's summary: "In Saudi Arabia, the price of electricity is whatever the government wants it to be."
Vision 2030, but practical
Riyadh's diversification drive has had mixed results in heavy manufacturing due to capital intensity and skills gaps. Petrochemicals are a bright spot. Mega-projects like The Line have been scaled back as the state refocuses on logistics, minerals, and religious tourism.
AI infrastructure, however, lines up with the country's core advantage: cheap, controllable energy at scale. The bet is straightforward: secure chips, build capacity, export compute.
The bottleneck: advanced chips
Saudi and the UAE have pushed for access to top-tier AI semiconductors. Approvals for Nvidia's newest Blackwell chips-reportedly including 18,000 units for Humain-remain a moving target amid concerns over China ties. NVIDIA Blackwell
Saudi executives publicly pledge not to share US tech with Chinese firms. Some in the region expect approvals to land eventually-Europe won't buy at scale, and China is constrained. But until chips ship, timelines slip.
Key risks to watch
- Grid build-out pace vs. data centre delivery schedules
- US export controls and licensing delays for AI accelerators
- ESG pressure on crude- and gas-fired compute
- Local talent pipeline for operations, cooling, and security
- Cooling technology fit for desert climates (water use, heat rejection)
What this means for governments, CIOs, and developers
- Run honest TCO models: compare Saudi siting vs. US/EU on energy, land, cooling, tax, and latency.
- Secure long-term energy contracts with indexed pricing and clear curtailment rules.
- Design for modularity: containerized builds, staged capacity, and flexible cooling (liquid/immersion).
- Balance ESG: pair fossil-sourced baseload with solar build-out and credible offsets or RECs.
- Plan for chip risk: multi-vendor GPU strategy, staggered deliveries, and workload portability.
- Legal and compliance: data residency, cross-border transfers, and third-party audits.
- Resilience: distributed siting across regions, on-site generation, and heatwave contingencies.
- Workforce: local training tracks for facilities, MLOps, and reliability engineering.
Bottom line
Cheap, controllable energy gives Saudi Arabia a real shot at becoming a compute supplier at global scale. If it solves chips, grid, and credibility in one motion, the economics are hard to ignore.
For teams building AI products, the smart move is optionality: price in a Saudi region for training or inference at scale, keep a domestic footprint for latency-sensitive workloads, and be ready to shift as approvals and prices move.
If you're aligning teams for this shift, these practical learning paths can help: AI courses by job role.
Your membership also unlocks:
