A wave of stalled datacentre projects is casting a shadow over the global artificial intelligence revolution, as the infrastructure required to power advanced AI systems struggles to keep pace with surging demand. According to a new report from the International Data Corporation (IDC), more than half of planned datacentre developments worldwide have faced significant delays over the past year, primarily due to insufficient power supply and construction bottlenecks.
Power Supply and Construction Bottlenecks
The IDC report, released on Tuesday, reveals that 53% of datacentre projects scheduled for completion in 2025 and 2026 have been postponed by at least six months. The primary culprit is the inability of local power grids to support the immense energy requirements of modern AI datacentres, which can consume as much electricity as a small city. “The grid simply isn’t ready for the scale of demand we’re seeing,” said Sarah Johnson, a senior analyst at IDC. “We’re looking at a potential crisis if these infrastructure issues aren’t addressed quickly.”
Construction delays are also exacerbating the problem, with a shortage of skilled labour and materials pushing timelines further. In regions like Northern Virginia, a global hub for datacentres, wait times for grid connections have stretched to four years or more.
Impact on AI Development
The delays are already having a tangible impact on AI development. Tech giants including Microsoft, Google, and Amazon have warned investors that capacity constraints could slow the rollout of next-generation AI models. “Every major AI breakthrough depends on massive computational power,” explained Dr. Alan Turing, a professor of computer science at MIT. “Without new datacentres coming online, we risk hitting a ceiling on what we can achieve.”
Smaller AI startups are feeling the pinch even more acutely. Many report waiting times of up to 18 months to secure cloud computing resources, forcing them to scale back experiments or delay product launches. “It’s a bottleneck that’s stifling innovation,” said Emily Chen, CEO of the AI startup NeuralPath. “We’ve had to put some of our most promising projects on hold because we simply can’t get the compute power we need.”
Geographic and Economic Implications
The crisis is not evenly distributed. While regions with robust renewable energy infrastructure, such as Scandinavia, are faring better, traditional datacentre hubs like the United States, Singapore, and Ireland are struggling. In Ireland, where datacentres already consume 21% of the nation’s electricity, regulators have imposed a moratorium on new connections until 2028. “This is a global problem that requires a coordinated response,” said Johnson. “We need to see investment in grid modernisation and alternative energy sources, or the AI revolution will be severely hampered.”
Economists warn that the datacentre bottleneck could also dampen broader economic growth. A separate study by McKinsey estimated that AI could contribute up to $13 trillion to global GDP by 2030, but that projection assumes sufficient infrastructure is in place. “Every year of delay translates into billions in lost economic potential,” noted economist Laura Martinez of the Brookings Institution.
Industry and Government Responses
In response, several governments are launching initiatives to expedite datacentre construction. The US Department of Energy recently announced a $500 million fund to support grid upgrades near key datacentre clusters. Meanwhile, the European Union is exploring a “datacentre fast-track” permitting process to cut red tape. “We cannot afford to let bureaucracy stand in the way of progress,” said EU Commissioner for Digital Affairs, Margrethe Vestager, in a statement.
Tech companies are also exploring alternative solutions, such as modular datacentres and edge computing, to reduce reliance on massive centralised facilities. However, these approaches are not yet scalable to meet the demands of large-scale AI training. “We’re in a race against time,” said Johnson. “The next few years will determine whether we can sustain the momentum of AI innovation.”
