Grid Delays Outweigh Demand in AI Data Centres
Imagine pouring millions into a shiny AI data centre, assuming it'll be powering algorithms faster than you can say 'digital transformation'. Yet, surprise—you might be stuck waiting for that grid power connection for a cozy seven to ten years in Europe. That's the cold reality served up by the latest IQ-EQ study, unveiled at the Sustainable Real Estate Forum's summit on November 3 in London.
Waiting Games in Energy Access
If you're thinking a 'power secured' asset means electricity will flow like water, think again. According to IQ-EQ, this definition can stretch like taffy—from requesting a connection to actually flipping the switch. Many investors fail to grasp the risk of their assets turning into tech ghosts haunting an energy-less shell.
The real kicker? Building the centre might take just two to three years, leaving the entire investment sitting on its hands waiting for grid love.
- Forecasted delay in grid connection: 7-10 years
- Construction time for facilities: 2-3 years
Shifting Grounds with Grid Policy
Norton Rose Fulbright's chapter shines a light on the evolving landscape of grid access. The game of musical chairs isn't fun when your seat is a multi-million-dollar project and the rules keep changing. British data centres hogged about 73 GW of demand by mid-2025, with rules now pushing projects to prove readiness or risk losing slots. In the Netherlands, sitting on contracted capacity too long could see it snatched away. The EU, meanwhile, is pushing to triple capacity but still lacks solid timelines.
Your prized queue position? Well, it’s now something you have to defend actively. And yeah, there's a heartbreaking gap between when your lease says a data centre should be ready and when it actually gets going.
Insurance and Investment Risks
According to Zurich Insurance, big projects cast big shadows on potential hiccups. The margin for error slims down as project scales balloon—the cost of a average data centre project insured by Zurich leapt from $150 million to $3 billion over five years. This underscores the importance of comprehensive early-stage risk and resilience checks.
Building a Robust Governance Framework
In response to these hurdles, the study proposes a multi-layered governance framework for AI infrastructure funds. This involves weaving energy commitments into fund documents and setting investor notification triggers when timelines go off track. Also crucial is distinguishing between assets that are 'energised' and those caught in the power queue when evaluating net asset value.
“AI demand is a technology thesis; converting it into durable cash flow is an infrastructure-delivery thesis,” says Peter Panayi of the Sustainable Real Estate Forum.
Real-World Examples
Consider the SINES Data Campus in Portugal—its SIN01 building stayed up and running despite a grid hiccup in April 2025, highlighting that well-planned, resilient energy infrastructure can withstand unexpected challenges.
Ultimately, the colossal challenge of balancing data centres with the energy transition is not a wild goose chase. It's feasible, but only if engineered with precision and governed meticulously. The sector needs more than just ambitions—sound strategies and robust frameworks must be at the core of turning today’s AI fantasies into tomorrow’s powerhouse realities.
