AI's Electricity Appetite Is Real: What the IEA's Numbers Mean for Gulf Grid Engineers
The IEA's latest data centre figures show global demand roughly doubling by 2030, and in the Gulf that growth is already being met with a nuclear regulator training staff on small modular reactors.
Every AI product launch comes with a quieter question attached: where does the electricity come from. The IEA has now put numbers on it, and the Gulf's answer is already visible.
The scale, in TWh
Data centres accounted for around 1.5% of the world's electricity consumption in 2024, or 415 terawatt-hours (TWh), according to the IEA's Energy and AI report. That consumption has been growing at around 12% a year since 2017, more than four times faster than total electricity consumption across all other sectors. The IEA projects it will more than double to around 945 TWh by 2030, just under 3% of global electricity consumption that year, with electricity demand from AI-optimised data centres specifically projected to more than quadruple over the same period. China and the United States are the most significant regions for data centre electricity consumption growth, accounting for nearly 80% of global growth to 2030, with the US alone seeing data centres responsible for nearly half of its total electricity demand growth to 2030.
The pace has not slowed since that report was published. A follow-up IEA analysis from April 2026 found data centre electricity demand grew 17% in 2025 alone, well outpacing the roughly 3% growth rate of global electricity demand overall, while tech company capital expenditure surged past $400 billion in 2025 and is expected to rise a further 75% in 2026.
Where the bottleneck actually is
The constraint isn't chips, it's grid capacity. The IEA's Energy and AI report finds that wait times for critical grid components such as transformers and cables have doubled in the past three years, and that around 20% of planned data centre projects could be at risk of delay because of this. The April 2026 follow-up adds that slow grid connections and stretched planning and regulatory systems are pushing some developers toward onsite gas generation instead of waiting for grid capacity, precisely the kind of workaround a region with spare generation capacity and an active grid-buildout programme is positioned to avoid.
The Gulf's specific answer: nuclear, considered explicitly for data centres
This is where the story becomes concrete for someone working in a UAE utility. In February 2026, the UAE's nuclear regulator, FANR, said explicitly that data centre demand is a driver behind considering further nuclear capacity. FANR Director General Christer Viktorsson said, "For data centres, one concept is to have a separate source of energy and it could be an SMR," referring to small modular reactors, and confirmed the regulator is "training staff in new technologies such as SMRs," while tracking roughly 80 to 90 different SMR designs currently on the market. No formal expansion proposal had been submitted as of that date, but the regulator's own framing ties a global AI-electricity trend directly to a live Gulf licensing question.
That connects to a wider pattern the IEA also flags: AI-related skills remain much less prevalent in the energy sector than in other industries, the same skills gap the IEA's World Energy Employment 2025 report quantifies at roughly 40% fewer AI-skilled workers, proportionally, in energy firms compared with tech, finance, education or media. A regulator training its own staff on SMR technology because of AI-driven demand is a direct, local instance of that global gap closing in real time, and it is closing through people, not just procurement.
What this actually changes for a grid career
None of this means data centres are about to dominate Gulf electricity demand the way they are starting to in parts of the US. It means the region's grid and nuclear planning now explicitly names AI-driven demand as a design input, and that the regulator overseeing it is retraining staff to match. A licensing and engineering environment that is actively absorbing a new reactor class for a new demand driver is, for an early-career engineer, a rare moment where the skill requirements of a job are still being defined rather than fixed.
Key sources
- IEA Energy and AI, Executive Summary: the IEA's core 2030 projections for data centre and AI electricity demand.
- IEA, "Data centre electricity use surged in 2025..." (April 2026): the most recent update on 2025 growth and grid bottlenecks.
- The National, "UAE's nuclear regulator 'ready' for capacity expansion" (February 2026): FANR's own statement linking data centre demand to SMR planning.
- IEA World Energy Employment 2025: the sector-wide AI-skills gap figure referenced above.
What this means for you
If you're a student
If you're studying nuclear, electrical, or grid engineering, look specifically for coursework or reading on small modular reactor design and licensing, not just conventional large-reactor systems. FANR naming SMRs as the concept under active study for data centre power means SMR literacy is a genuinely forward-looking specialisation in this market right now, not a speculative one.
If you're early-career
If you work in grid planning, protection, or SCADA at a utility or the regulator, ask directly whether your organisation is involved in the SMR technology evaluation FANR has described. Volunteering for that work, even in a support capacity, puts you inside a live regulatory and engineering process before it has settled into a fixed set of standardised roles.