AI & TechArtificial IntelligenceMENA Tech SceneNewswireTechnology

UAE Splits $30bn AI Data Center After Iran Named It a Target

▼ Summary

– The UAE is redesigning its massive Stargate AI infrastructure project from a single large campus into a distributed network of smaller sites to mitigate security risks.
– This strategic shift follows Iranian threats and attacks on US facilities in the region, prompting the adoption of military-grade protections for data centers.
– New security measures include underground facilities, mountain-embedded workloads, air defenses, and blast-resistant construction to protect against missiles and drones.
– The project is led by Emirati group G42 in partnership with OpenAI, Oracle, and SoftBank, aiming for a five-gigawatt capacity across multiple phases.
– Security concerns in the Gulf have become a significant hurdle for AI expansion, similar to regulatory and energy issues that previously stalled the UK phase.

The United Arab Emirates has fundamentally altered its strategy for the Stargate project, the largest artificial intelligence infrastructure initiative in the Gulf region. According to an exclusive report by Reuters, the original vision of a single, massive campus spanning ten square miles of desert near Abu Dhabi has been abandoned. Instead, planners are now designing a distributed network of smaller data centers scattered across the country. This shift is driven by a stark security reality: consolidating such a critical asset into one location makes it an easy target.

The decision follows heightened tensions after Iran explicitly identified Stargate as a military objective. In April, Iranian armed forces released a video naming the site and displaying a map that placed it adjacent to Al Dhafra Air Base, a key hub for American aircraft. This clear signal prompted engineers to adopt protection measures typically reserved for military installations rather than commercial tech facilities. These enhancements include burying facilities underground, constructing sensitive workloads within mountains, installing advanced air defense systems to intercept drones and missiles, using blast-resistant concrete, and creating redundant power and cooling systems.

Security Over Efficiency

These engineering changes represent a significant departure from standard data center practices. Building for missile resistance adds substantial cost and complexity to projects designed primarily to run graphics processing units. The first phase of the UAE project, valued at $30 billion, involves a one-gigawatt cluster with 200 megawatts expected to be operational this year. The long-term plan aims for five gigawatts of capacity. The venture is led by G42, the Emirati AI group controlled by Sheikh Tahnoun bin Zayed al Nahyan, in partnership with OpenAI, Oracle, SoftBank, Nvidia, and Cisco.

While the first phase continues with modifications, the project faces scrutiny similar to other international expansions. OpenAI recently paused the UK arm of Stargate due to energy costs and regulatory hurdles, marking the second national project in the program to encounter non-demand-related obstacles. The threat landscape in the Gulf has complicated the region’s pitch to the AI industry. For years, Gulf states attracted investment by offering cheap energy, capital, and space. However, recent conflicts have transformed the geographic location itself into a primary risk factor.

Strategic Realignment

Abu Dhabi has previously marketed its financial center as holding $100 billion in AI capital. Redesigning the flagship site to withstand missile attacks undermines that narrative of stability. Beyond physical security, G42 is navigating complex geopolitical pressures. Over the past two years, the company has dismantled its relationship with Huawei and opened its facilities to American oversight to maintain access to advanced semiconductors. Reports suggest that majority US ownership may follow soon.

Simultaneously, G42 is acquiring capacity in safer jurisdictions, such as a converted office building in Minneapolis. Spreading operations abroad while decentralizing domestic sites appears to be a consistent strategy to mitigate risk. However, dispersing a gigawatt of computing power across multiple locations introduces technical challenges. Large AI training models require chips to be physically close because the cabling between them is integral to performance. Splitting clusters geographically can degrade efficiency.

The full financial and schedule impact of these hardening measures remains unclear. No public figures have been released regarding the additional costs or delays. Until the revised masterplan is finalized, stakeholders outside the project cannot accurately assess how these security-driven changes will affect the timeline or budget of this landmark infrastructure effort.

(Source: The Next Web)

Topics

ai infrastructure security 98% geopolitical risk in gulf 95% stargate project evolution 92% uae tech strategy 88% global ai expansion challenges 85%
Show More