UAE Could Export AI Compute Rather Than Energy, G42 Chief Says

What happened

G42 chief executive Peng Xiao has said the UAE can reposition itself as a global exporter of intelligence rather than energy, by turning its energy resources into computing power that could supply AI services far beyond its borders. He made the remarks at the Dubai AI Festival on Monday, as reported by The National on 26 October 2026.

Xiao described the ambition as moving from being an energy exporter to being an intelligence exporter. He argued that the UAE's access to reliable energy, combined with its speed in delivering computing infrastructure, could help it meet international demand for AI services.

G42 is developing a five-gigawatt AI campus in Abu Dhabi. The campus includes the one-gigawatt Stargate UAE project, which involves partners OpenAI, Oracle, Nvidia, Cisco and SoftBank. Xiao said that many American partners are now visiting the country to secure access to compute capacity, which he attributed to the UAE's current ability to deliver it.

Why it matters

The story is less about energy policy than about where AI value is created and sold. If compute capacity becomes the product, the UAE's proposition shifts from hosting AI projects to supplying the infrastructure that other organisations build their services on. For leaders in digital transformation, that changes the question from whether to adopt AI to where their AI workloads will run, and on what terms.

For service design, the practical consequence is that the performance of AI products increasingly depends on infrastructure decisions made upstream. Capacity, speed of deployment and reliability shape what end users actually experience, including response times, consistency and availability. Organisations serving customers with AI will need to treat these dependencies as part of the service itself.

By the numbers

• 5 GW: the planned scale of G42's AI campus in Abu Dhabi.
• 1 GW: the capacity of the Stargate UAE project within that campus.
• 5 partners: OpenAI, Oracle, Nvidia, Cisco and SoftBank, named in the Stargate UAE project.
• 26 October 2026: the date of the reporting on Xiao's remarks.

The Renascence take

Most readers will focus on the headline ambition, but the customer-facing question is whether capacity translates into dependable service. Announcements about gigawatts are easy to make; the experience of an AI product is shaped by latency, uptime and how consistently it behaves on a bad day.

Customers do not judge AI by the size of the data centre behind it; they judge it by whether the answer arrives quickly and the service works when demand spikes. Trust in AI services is built on reliability and transparency, so the infrastructure story is really a service-design story. Operators should map every customer promise back to the compute, power and vendor dependencies that deliver it, and design visible fallbacks before they design marketing around capacity.