The United Arab Emirates is adjusting its ambitious plan for a major artificial intelligence (AI) data center known as the UAE-US AI Campus. Originally envisioned as a single, massive facility covering 26 square kilometers near Abu Dhabi, the project is now being restructured into a network of smaller data centers spread across the country. Some of these installations will be buried underground, and sensitive data will be stored in the mountainous regions of the northeast. This change follows security concerns and technical challenges, including threats from Iran and damage to existing data centers in the region. The shift in strategy comes in the wake of missile and drone attacks by Iran in March, which targeted several Gulf countries where U.S. military forces are stationed. These attacks damaged Amazon Web Services data centers in the UAE and another in Bahrain. In April, Iran's military released a video that explicitly identified the UAE-US AI Campus as a potential target, revealing its previously secret location. As a result, the UAE is now incorporating advanced security measures into its revised plan, including air defense systems to intercept drones and missiles, blast-resistant concrete, and backup power and cooling systems. The UAE-US AI Campus is being developed by G42, an AI company controlled by Sheikh Tahnoun bin Zayed al Nahyan. Major technology firms such as NVIDIA, Oracle, Cisco, and SoftBank are also involved in the project, which is part of a larger initiative called Stargate. This initiative aims to provide OpenAI, the organization behind the popular AI model GPT, with substantial computing power. The first phase of the project is estimated to cost $30 billion and aims to achieve 1 gigawatt of computing power, with 200 megawatts expected to be operational this year. The ultimate goal is to reach 5 gigawatts of computing power. However, spreading the AI infrastructure across multiple locations presents significant technical challenges. Large AI models require computing chips to be as close together as possible to minimize delays and maximize performance. Distributing the infrastructure increases latency, reduces efficiency, and may require changes to how AI models are structured. This could result in longer training times and less performance for the same budget. Additionally, the project faces energy-related challenges similar to those encountered in the UK, where the Stargate project was temporarily paused. Meeting the power requirements for 1 gigawatt involves substantial electrical infrastructure, including high-voltage connections and large supply contracts. Cooling systems must be highly advanced to manage heat, and the project must ensure high availability with backup power and network systems. Local factors such as energy availability, water supply, permits, and supply chain issues can also affect the project's timeline and costs. The revised plan includes a multi-site architecture with built-in redundancies and security features. This means ensuring strong interconnections between sites, using redundant fiber optic cables, alternative network routes, and data replication. Physical resilience, like blast-resistant materials and drone protection, must be combined with IT continuity measures such as backup power, cooling systems, and failover procedures. However, this distribution makes operations more complex and can reduce the performance of AI clusters that rely on low latency. The UAE's decision to revise the project represents a balance between ensuring the facility's security and minimizing costs, while also maintaining the efficiency of its computing power.