HUMAIN & DataVolt Begin 100MW AI Data Center Construction at NEOM Oxagon

Breaking Ground on a New Era of Compute

The race for frontier AI supremacy has shifted unequivocally from merely developing foundational software models to establishing the colossal physical infrastructure required to sustain them. In a monumental step for the Middle East’s technological ambitions, HUMAIN, a leading artificial intelligence company backed by the Public Investment Fund (PIF), and DataVolt, a premier sustainable digital infrastructure operator, have officially transitioned their joint 100MW AI data center project from the strategic planning phase to active, large-scale construction at Oxagon. Oxagon is the advanced manufacturing and innovation city dynamically evolving within NEOM, Saudi Arabia.

This groundbreaking ceremony marks the commencement of what is projected to become one of the most concentrated and advanced hubs of high-performance compute (HPC) globally. As artificial intelligence models scale exponentially—moving from billions to trillions of parameters—demanding unparalleled power and cooling capacities, traditional data centers are rapidly becoming technologically obsolete. The HUMAIN-DataVolt facility represents a fundamental paradigm shift, architected entirely from the ground up to support the extreme densities of next-generation AI accelerators.

The 100MW facility currently under construction is merely the vanguard of a much larger, highly ambitious vision. It constitutes the initial, critical tranche of a planned 360MW first phase, which will eventually culminate in DataVolt’s massive 1.5-gigawatt (GW) AI data center campus sprawling across the Oxagon landscape. Scheduled to come online and begin processing workloads by 2028, this inaugural 100MW installation is a critical pillar in Saudi Arabia's Vision 2030. It is strategically engineered to establish the Kingdom not just as a wealthy consumer of AI technologies, but as a sovereign powerhouse of global AI compute capacity.

Architectural Engineering for High-Density AI Workloads

The architectural blueprint of the HUMAIN-DataVolt center at NEOM deviates significantly from conventional cloud infrastructure designs of the past decade. Frontier AI training clusters, such as those utilizing the latest architectures detailed in the comprehensive NVIDIA Blackwell Ultra GB300 vs H200 analysis, generate unprecedented and highly concentrated thermal footprints. Server racks that historically consumed a modest 10-15 kilowatts (kW) of power are now being rapidly superseded by specialized AI racks demanding upwards of 120kW to 150kW.

To accommodate this extreme density, the facility incorporates a heavily reinforced structural design and specialized spatial configurations to house massive GPU clusters safely. The building must support the immense weight of liquid cooling manifolds and the densely packed hardware without compromising on operational efficiency, maintenance accessibility, or physical safety protocols.

Direct-to-Chip and Immersion Liquid Cooling Technologies

Traditional air cooling is mathematically and thermodynamically insufficient for modern, high-intensity AI workloads. Recognizing this physical limitation early in the design phase, the Oxagon facility is deploying state-of-the-art high-performance liquid cooling systems. In a strategic collaboration with industry leaders like LG Electronics and specialized advanced chiller technology partners, DataVolt is implementing a robust hybrid cooling strategy.

This intricate system involves both Direct-to-Chip (D2C) cooling, where precision-engineered cold plates are affixed directly to the GPUs and CPUs to rapidly wick away heat via a circulating, specially formulated fluid, and built-in architectural provisions for future immersion cooling, where entire hardware chassis are safely submerged in a non-conductive dielectric fluid. This multifaceted approach not only prevents hardware thermal throttling during multi-month, high-demand training runs but also drastically reduces the facility's Power Usage Effectiveness (PUE) ratio, targeting an industry-leading PUE of near 1.05.

Sustainable Power: 100% Renewable Energy Integration

One of the most profound bottlenecks facing AI companies globally today is the availability of abundant, clean power. As highlighted by the massive energy demands discussed in the NVIDIA Reports $96.2B Q2 Revenue financial breakdown, raw compute power is intrinsically linked to electrical power generation.

The strategic geographical placement of this massive data center at NEOM is no coincidence. It is intricately designed to be fully integrated with NEOM’s bespoke 100% renewable energy grid, drawing immense amounts of power entirely from expansive, next-generation solar arrays and wind farms located within the surrounding region. This crucial integration completely eliminates the massive carbon footprint typically associated with training trillion-parameter frontier models, offering a highly compelling value proposition for global AI developers who are facing mounting international regulatory pressure regarding the environmental sustainability of their operations.

Saudi Arabia's Vision 2030 and Sovereign Compute Infrastructure

The initiation of this massive construction project is a highly tangible manifestation of Saudi Arabia’s sweeping Vision 2030 initiative. The Kingdom is aggressively and strategically pivoting from a historically hydrocarbon-based economy to a dynamic, knowledge and technology-driven ecosystem. By establishing world-class sovereign compute infrastructure, Saudi Arabia ensures that it firmly retains control over its critical data, intellectual property, and its overall AI developmental trajectory.

The Emerging Geopolitics of AI Infrastructure

In the current, highly competitive geopolitical climate, AI compute capacity is increasingly viewed as a critical national resource, akin to strategic petroleum reserves or rare earth metals. Sovereign data centers meticulously mitigate the substantial risks associated with over-reliance on foreign cloud providers, safeguarding highly sensitive national data and proprietary intellectual property. The HUMAIN-DataVolt partnership ensures that the Kingdom possesses the requisite "heavy machinery" to independently train localized foundational models and broadly support regional enterprise AI adoption.

Furthermore, this cutting-edge infrastructure naturally positions Saudi Arabia as an incredibly attractive destination for international AI startups and established global tech giants. By actively offering access to abundant, green power and cutting-edge cooling tech that is often scarce elsewhere, the Oxagon campus aims to draw top-tier global talent and substantial foreign investment, fostering a vibrant, localized ecosystem of technological innovation.

The Critical Intersection of Infrastructure and Cybersecurity

As the physical scale of AI infrastructure expands globally, so too does the potential attack surface for malicious actors. High-value targets like a state-of-the-art 100MW AI data center require absolutely unprecedented security measures. The robust physical security of the Oxagon facility is tightly integrated with and complemented by advanced logical security frameworks.

Protecting the pristine integrity of the massive datasets and the highly valuable models being trained is paramount. This necessitates robust, AI-native defenses against sophisticated, AI-driven threats, similar to the advanced solutions currently being pioneered in the cybersecurity sector, such as those comprehensively covered in the CrowdStrike Launches Falcon IQ & Expands QuiltWorks announcement.

Moreover, as autonomous, goal-seeking AI agents become significantly more prevalent in enterprise environments, the underlying infrastructure hosting them must enforce rigorous operational boundaries. Sophisticated techniques for monitoring and safely containing these intelligent systems, as explored in the detailed guide on how to Build an AI Agent Sandbox Escape Detection Workflow, will be crucial for the safe, continuous operation of the frontier models housed within the secure HUMAIN-DataVolt center.

Comprehensive Comparative Analysis: Oxagon vs. Traditional Data Centers

To truly grasp the immense magnitude and technical leap of the HUMAIN-DataVolt initiative, it is essential to compare its specifications directly against standard enterprise data centers currently in operation.

Data compiled from official DataVolt architectural projections and current industry standards for AI-optimized infrastructure construction.

The Broad Economic and Technological Ripple Effects

The multi-year construction of this massive facility will generate significant, sustained economic momentum within NEOM and the broader Saudi economy. Beyond the immediate creation of thousands of high-tech construction and specialized engineering jobs, the long-term operational phase will require a highly specialized, permanent workforce of data center technicians, liquid cooling mechanics, and advanced AI systems engineers.

Accelerating Regional and Global AI Research

The guaranteed availability of massive, localized compute power will drastically accelerate academic and commercial research initiatives within the region. Regional universities, research institutions, and newly formed startups will have unprecedented access to the colossal compute resources necessary to train complex, highly nuanced models tailored to local Arabic languages, specific cultural contexts, and highly specialized industrial needs—ranging from autonomous shipping logistics at the cutting-edge Oxagon port to advanced desalination and climate modeling.

As AI models rapidly evolve in capability, the vital importance of robust safety and alignment protocols becomes incredibly critical. The massive infrastructure at Oxagon will undoubtedly support the rigorous testing and deployment of advanced AI safety frameworks at an unprecedented scale, akin to those being developed by leading labs globally, such as the mechanisms detailed in the Anthropic Launches Claude Agent Guardrails v2 framework, meticulously ensuring that the AI systems trained on this powerful hardware operate strictly within defined, secure, and ethical boundaries.

Looking Ahead: The Ambitious Road to 2028 and Beyond

The project timeline for the HUMAIN and DataVolt joint venture is notably aggressive. With the initial, massive 100MW tranche targeted for a full operational launch by 2028, the sprawling construction teams at Oxagon face the monumental, unprecedented task of perfectly integrating bleeding-edge, unproven cooling tech with massive, utility-scale electrical infrastructure in a highly challenging desert environment.

However, the successful, on-time execution of this monumental project will mark a definitive watershed moment in technological history. It will undeniably prove the commercial and technical viability of powering frontier AI operations entirely on renewable energy at a massive global scale, a critical, necessary milestone for the global tech industry's long-term sustainability goals.

As the deep physical foundation of the next phase of the AI revolution is literally poured into the sands of NEOM, the HUMAIN-DataVolt data center stands as a towering testament to the sheer scale of financial and infrastructural investment required to participate meaningfully in the next era of computing. It is not merely a highly advanced building; it is the physical engine room for the next generation of artificial intelligence, strategically positioned at the geographic and technological crossroads of the world.

By Deepak Bagada, CEO at SaaSNext & Principal AI Architect.
Last tested & verified: August 2026 with Python 3.12, Node v22, and latest framework releases.