How Barakah nuclear plant gave UAE a head start in powering AI infrastructure

Barakah’s four reactors produce around 40 terawatt-hours of clean electricity annually, equivalent to about 25 per cent of the country's electricity needs

  • PUBLISHED: Fri 12 Jun 2026, 4:45 PM

In 2007, a study commissioned by the Abu Dhabi government set out to answer a straightforward but critical question — how much electricity would the UAE need to power its future?

The projections took into account population growth, expanding industrial activity and major infrastructure developments. Aluminium smelters were coming online, desalination plants were expanding, and cities were growing at a pace few countries had experienced.

The study found that by 2020, the UAE would need around 40,000 megawatts of power. Existing generation capacity was expected to reach only about half that figure.

Stay up to date with the latest news. Follow KT on WhatsApp Channels.

“There was a demand-side management study that extrapolated population growth from 2007 to 2020, and the conclusion was that you would need 40,000 megawatts by then,” Talal Al Kaissi, CEO of Core42 and Acting Chief Global Affairs Officer at G42, told Khaleej Times.

“With the production capacity available at the time, you were only going to reach around 20,000. There was no line of sight to anything called AI that would have given us the advantage we have today. Population growth and industrial capacity coming online. That was the calculation," he added.

'We were ready when the next demand arrived'

Natural gas was one option. Solar was another, though panel efficiency and scale at the time made it harder to meet the projected demand alone. Nuclear entered the conversation as an answer to a serious industrial problem, and the idea moved with speed.

“It came out of an idea like that, and leadership embraced it, just like the Mars mission,” Al Kaissi said. “Because we started that power journey seriously in 2010, we were ready when the next demand arrived. That is exactly how this country has always operated — leadership, foresight, execution.”

Al Kaissi was directly involved in that journey, working out of the UAE Embassy in Washington and supporting efforts to secure the legal and diplomatic framework behind what became the Barakah nuclear programme, including the 123 Agreement that enabled peaceful nuclear cooperation.

Construction began around 2010. Today, Barakah’s four reactors produce around 40 terawatt-hours of clean electricity annually, equivalent to about 25 per cent of the UAE’s electricity needs. The decision that produced them had nothing to do with machine learning. But the infrastructure it created is now directly relevant to a race few had conceived in 2007.

That same period also showed another early signal of the UAE’s technology direction. In 2007, Mubadala invested in AMD. A year later came GlobalFoundries. Those moves were not about today’s generative AI boom, but they formed part of a wider diversification journey into the technologies and industries that would later sit closer to the foundations of artificial intelligence.

Over the past few years, the world has reached the brutal realisation that AI is not only software, data science, or engineering. Its main enabler is power and electrons. The International Energy Agency projects global data centre electricity consumption will more than double to around 945 terawatt-hours by 2030, with data-centre demand growing far faster than overall electricity consumption. Governments and companies racing to build AI infrastructure are running into grid constraints, permitting delays, and the long lead times required to build new generation capacity from scratch.

Luckily, the UAE is not on the same queue.

“If we had to start from scratch on nuclear when the five-gigawatt AI campus conversation took place, it would have taken us more than 10 years to build the power needed to make that even a viable option,” Al Kaissi said. “Because we started that power journey early, we were ready. If you look at other countries getting into AI infrastructure right now, many of them are discovering that they lack exactly this kind of foundational capacity. Power is the constraint nobody budgeted for.”

How did this come to be?

The technology trigger came in 2016 when Google’s DeepMind built a system called AlphaGo and matched it against the world champion of Go, a game of such combinatorial complexity that it had long been considered beyond machine reach. The system played a move in the fourth match so counterintuitive that expert commentators assumed it was a mistake. It won four games to one.

What the result showed the world was that machine intelligence had crossed a threshold, moving from narrow pattern recognition into something capable of reshaping healthcare, education, government and industry.

UAE leadership recognised that shift before most governments had formed a view on it.

“In the 2017 timeframe, our leadership saw what happened with Google DeepMind and understood the strategic value of AI at a very early stage,” Al Kaissi said. “It became evident that if AI can leap off the board and into the real world like that, it stops being a technology story and becomes a national capability story. That insight is what set everything that followed in motion.”

Around that period, the UAE appointed the world’s first Minister of State for Artificial Intelligence. Research institutions were formed around machine learning, computer vision and natural language processing. National AI companies emerged, Mohamed bin Zayed University of Artificial Intelligence was established, and government and industry began adopting AI in ways that created a wider ecosystem rather than isolated projects.

In 2007, no one asked on the capacity for advanced tech and AI. It was how to power the next stage of national growth. Nearly two decades later, the answer is seems to be helping power both.