Saudi Arabia’s Multi-GW AI Factory Plan: How HUMAIN and DataVolt Are Building a National Compute Backbone

Just a few years ago, Saudi Arabia’s digital footprint was mostly defined by its smart-city dreams in NEOM and its rapid cloud adoption. But something has changed—quietly, and almost suddenly.

Riyadh is no longer talking about innovation; it is building the physical backbone of global AI itself.

What began as scattered data-center announcements has evolved into something far more ambitious:
a national plan to turn Saudi Arabia into one of the world’s largest producers of AI compute.

And the catalyst is a new partnership—HUMAIN and DataVolt—whose multi-gigawatt AI factory blueprint is unlike anything the world has seen outside the U.S. and China.

The Large-Scale Push Toward a National “Next-Generation AI Factory”

In early November 2025, a quiet but highly meaningful announcement came out of Riyadh.

Saudi Arabia’s PIF-owned AI company HUMAIN and the data-center developer DataVolt officially declared that they will build a multi-gigawatt AI data-center pipeline across the kingdom.

What they announced is not about building “a few big data centers.”

This pipeline represents a plan to spread several gigawatts of AI-dedicated compute infrastructure across Riyadh, Dammam, NEOM, and additional strategic regions—effectively a nation-level AI factory network.

A separate 1.5GW net-zero AI Factory Campus is already under construction in NEOM’s Oxagon district, and the new HUMAIN–DataVolt partnership ties these elements together as a unified national blueprint.

HUMAIN: A National AI Company Aiming for a 6–7GW Compute Base

HUMAIN was launched in May 2025 by Crown Prince Mohammed bin Salman.

It is structured as a PIF-owned national AI company, with the Crown Prince serving as Chairman, and its vision is far larger than a conventional startup.

HUMAIN is designed to function as Saudi Arabia’s national AI platform provider.

HUMAIN as being responsible for:

• Building next-generation data-center and cloud infrastructure

• Operating large-scale GPU clusters and AI compute

• Developing large language models, including Arabic LLMs

• Providing AI solutions for government, industries, and smart-city environments

HUMAIN plans to build up to 6.6GW of data-center capacity by 2030, with an implied investment that could reach USD 77 billion.

There is even reporting that HUMAIN aims to run as much as 7% of global AI workloads inside Saudi Arabia.

China’s Physical AI Ecosystem Explained: Key Players, 2025–2030 Roadmap, and Global Implications

Massive GPU Procurement: NVIDIA, AMD, and More

HUMAIN’s scale is even clearer when examining its GPU strategy:

• NVIDIA agreement:
  Long-term procurement of approximately 180,000 GPUs, starting with an initial shipment of 18,000 Blackwell (GB300) GPUs.

• AMD agreement:
  A USD 10 billion collaboration focused on AI chips and 500MW-class infrastructure build-outs.

• SDAIA–NVIDIA: Sovereign AI Factory:
  A parallel initiative planning 5,000 Blackwell GPUs and a 500MW-class national AI facility for government use.

These GPUs will power a 500MW AI-optimized data-center line, designed not for traditional workloads but for massive-scale AI training and inference.

Thus, HUMAIN operates simultaneously as:

• a national AI company

• a hyperscale cloud and compute operator

• a massive GPU buyer

• and a national AI infrastructure integrator

This combination makes HUMAIN remarkably unique among AI players worldwide.

DataVolt: The Developer Behind Saudi Arabia’s AI Factory Campuses

DataVolt, founded in 2023 under Vision Invest, is not a typical data-center company.

Its focus lies in AI-centered campuses, where power, cooling, and sustainability are optimized for high-density compute clusters.

DataVolt’s Major Projects

In the HUMAIN partnership, DataVolt is responsible for:

• Site development and campus planning

• Power infrastructure, including high-capacity substations

• High-density cooling, including liquid cooling

• Net-zero and ESG-driven data-center designs

Put simply, DataVolt is the physical builder of Saudi Arabia’s AI factories.

The “Multi-Gigawatt AI Factory Pipeline”: What It Actually Looks Like

In their November 6–7 announcement, HUMAIN and DataVolt stated that they will jointly build a multi-GW pipeline of AI-oriented data centers throughout Saudi Arabia.

This pipeline includes HUMAIN-owned builds, DataVolt-developed projects, NEOM’s mega-campus, and other leased capacity.

Industry estimates place the full pipeline at 6.4–6.6GW over the next decade.

Phase 1: 100MW-Class AI Facilities (Target: 2026)

The first phase consists of two 100MW facilities:

• Riyadh – 100MW AI data center

• Dammam – 100MW AI data center

Both aim for operational readiness in mid-2026.

These facilities will receive the initial batch of 18,000 NVIDIA Blackwell GPUs already confirmed for shipment.

A 100MW AI-optimized facility supports:

• thousands of high-density GPU racks

• MW-scale thermal loads

• strong optical interconnects for AI training

• liquid and hybrid cooling

• large-scale data-fabric topologies

Unlike conventional IT data centers, these are designed from day one as AI factories, not multi-purpose IT sites.

Phase 2: NEOM Oxagon’s 1.5GW Net-Zero AI Factory

The next phase is even larger:  a 1.5GW AI mega-campus in NEOM Oxagon.

To put 1.5GW into perspective:

The NEOM campus includes:

• multiple hyperscale compute halls

• renewable-integrated or dedicated power

• advanced heat-recovery and liquid cooling

• logistics and robotics-ready design

• optical networking optimized for global AI workloads

Phase-1 investment alone is USD 5 billion, with full commissioning targeted for 2028.

Phase 3: A National Multi-Hub AI Compute Network

When combined, HUMAIN–DataVolt’s pipeline forms a three-hub compute architecture:

These hubs align with:

• The Sovereign AI Factory (SDAIA–NVIDIA)

• The USD 1B Dammam AI Hub (Google Cloud + HUMAIN)

• Future AWS, Microsoft, and OpenAI regional deployments

The overall picture becomes unmistakable:

Saudi Arabia is designing the entire kingdom as one distributed AI factory network.

Why “Multi-GW AI Factories”? The Core Logic: Power, Cooling, and Scale

As AI models grow, compute demand is no longer incremental—it is exponential.
To run next-generation FP8/FP4 models and frontier multimodal architectures, countries need:

• massive GPU installations

• ultra-dense power (MW per hall)

• advanced liquid-cooling systems

• a reliable, low-cost energy backbone

Saudi Arabia Holds Three Major Advantages

1. Energy Cost and Abundance
   Cheap and stable long-term electricity from oil, gas, and rapidly rising renewable generation.
   NEOM’s net-zero campus will integrate large-scale renewables to reduce carbon and energy costs.

2. Ability to Build “AI-First” Power Infrastructure
   Rather than forcing data centers into already strained urban grids, Saudi Arabia can build new, AI-optimized grids, substations, and cooling systems from scratch.

3. Policy Alignment under Vision 2030
   AI infrastructure is a central pillar of the National Data & AI Strategy (NSDAI), accelerating land, permitting, and power approvals.

In other words, HUMAIN–DataVolt’s pipeline is not merely about constructing data centers—it represents:

A shift from being an energy exporter to becoming an AI compute exporter.

Saudi Arabia is effectively turning electricity into compute and compute into economic value.

Strategic Meaning of the HUMAIN–DataVolt Pipeline

1) The emergence of “compute as national infrastructure”

Saudi Arabia views compute like it views:

• roads

• ports

• industrial zones

• power grids

A shift is happening:  AI compute is becoming a sovereign asset.

2) Compete with hyperscalers? Or host them?

HUMAIN is positioned both as:

• a domestic AI hyperscaler, and

• a potential host for global hyperscalers (AWS, Google Cloud, OpenAI, xAI).

Saudi Arabia could become:

• the Middle East’s primary compute hub, and

• a global overflow zone for AI training workloads.

3) A new force in global GPU demand

Until now, GPU allocation was driven by:

• U.S. Big Tech

• China’s AI ecosystems

• global hyperscalers

Now, Saudi Arabia has emerged as a fourth major bloc with nation-scale purchasing power.

4) A template for other energy-rich states

Saudi’s model could be replicated by nations with cheap, abundant energy:

• UAE

• Qatar

• Kazakhstan

• Algeria

• parts of Africa

This may become the post-oil economic model for many resource nations.

Japan’s Latency Gamble: How IOWN APN Is Rewriting AI Infrastructure

Strategic Questions for Global Stakeholders

Before moving to the strategic questions, AI Strategica asked a group of senior AI-industry stakeholders a simple but essential question: “Given everything unfolding in Saudi Arabia’s AI infrastructure push, what should the world’s technology and policy leaders be thinking about right now?”

The insights we gathered shaped the set of questions below—questions that global executives, policymakers, hyperscalers, energy companies, and semiconductor leaders are now actively wrestling with.

Below are the key strategic questions we captured for this CoreBrief.  Please Contact@AIStrategica.com for report purchase.

HUMAIN–DataVolt’s project forces the global AI and semiconductor ecosystem to consider several critical questions:

1. Is compute now a form of national infrastructure?
   Should governments own their own compute factories?

2. Can a HUMAIN-style model be replicated without cheap energy and sovereign wealth funds?

3. How should hyperscalers respond?
   Compete with Saudi Arabia—or partner with it?

4. How does one manage GPU procurement in a world where nation-states buy at scale?

5. What does “compute sovereignty” mean in 2030?
   Should a country rely on foreign AI megafactories?

6. Is this the beginning of a global race for national AI factories?
   If Saudi Arabia succeeds, who follows next?

The full InDepth is available exclusively to AI Strategica clients and subscribers.

➡Request Access
Contact us at Contact@AIStrategica.com to receive pricing, subscription options, and a sample excerpt.