Imagine a world where AI powers everything from virtual assistants to self-driving cars – but the infrastructure needed to support it is crumbling. A new report from the Uptime Institute reveals that nearly half of the globe's largest datacentre projects are facing significant delays or cancellations, threatening to throttle the rapid expansion of artificial intelligence (AI). These massive facilities, needed for training and operating advanced AI models, are struggling to keep pace with demand, prompting warnings that the global AI revolution is at risk.
The sheer scale of energy required by these datacentres is a major concern. Google has already admitted its cloud business, reliant on datacentres for AI services, is 'compute-constrained'. According to the Uptime Institute, even with cancellations and delays, there will still be an unprecedented increase in power demand over the next five years. Last year alone, planned projects could consume 1.3% of the world's projected electricity usage for 2025 – a near-doubling of current datacentre demand.
Several factors are contributing to these widespread issues. Jay Dietrich, a research director at Uptime, points to inexperienced developers and a lack of committed tenants as major concerns. The enormous energy and water consumption of individual projects is also causing problems, as is the concentration of facilities in 'datacentre corridors'. Global supply chain issues, particularly concerning chips for powerful machines, are exacerbating delays – with the Uptime Institute suggesting the global supply chain simply cannot support the projected timeline for the current level of projects.
Local opposition is also playing a significant role. In the US, a major project known as the Prince William Digital Gateway in Virginia is now in doubt following a court ruling and the withdrawal of a key backer. Opponents cited concerns about the project's proximity to a Civil War battlefield, arguing that 'monstrous datacentres' would mar the historic site.
The UK, while not the primary location for these 'mega-gigawatt' datacentres, is deeply integrated into the global digital economy and reliant on AI advancements. Delays in essential infrastructure abroad could impact the availability and cost of AI services for British businesses and consumers – with implications that could have economic repercussions for the country.
The escalating demand for power from these facilities is also raising concerns about the resilience of existing electricity grids. As the world grapples with the challenges of supporting a rapidly growing datacentre sector, one thing is clear: the infrastructure needed to support AI must be built quickly – or risk slowing down its global development.