Latest AI News

AI Data Centre Buildout Faces Toughest Challenge Yet: What UK SMEs Need to Know

The global AI infrastructure buildout, critical for advancing artificial intelligence, is facing unprecedented challenges in 2026, primarily due to local opposition, power constraints, and supply chain bottlenecks. This could impact the accessibility and cost of AI for UK SMEs.

Published 4 August 2026 · 3 min read

The ambitious global buildout of AI data centres, vital for the continued advancement and accessibility of artificial intelligence, is encountering its most significant hurdles to date in 2026. Local opposition, escalating energy demands, and persistent supply chain issues are collectively slowing down the rapid expansion needed to support the AI revolution. This emerging bottleneck has critical implications for UK SMEs, affecting everything from operational costs to the pace of AI adoption.

What Happened: The Infrastructure Crunch

Since OpenAI's release of ChatGPT 3.5 in November 2022 ignited the current AI race, hundreds of billions of pounds have been committed globally to construct the necessary AI infrastructure. However, this momentum has hit a significant snag in 2026. A primary challenge stems from widespread local opposition in the US, where citizens are increasingly pressuring officials to restrict or ban new data centre construction. Over 500 counties and municipalities in the US now actively limit or block data centre builds, with nearly 190 new restrictions enacted since June 2026 alone. This backlash is driven by concerns over energy consumption, noise pollution, and environmental impact, leading to delays or outright cancellations of projects worth billions. In the first quarter of 2026 alone, 75 data centre projects valued at approximately $130 billion were delayed or rejected due to local opposition. Compounding this, the energy requirements for AI data centres are skyrocketing. Gartner forecasts global data centre electricity consumption to reach 565 terawatt-hours (TWh) in 2026, a 26% increase from 2025. By 2027, AI-optimised servers are expected to consume more electricity than conventional servers, and by 2030, total data centre power consumption could exceed 1,200 TWh. This surge is putting immense strain on existing power grids, with some estimates suggesting that data centres could consume over 1,000 TWh by 2026, making them the fifth-largest energy consumer globally if they were a country. This energy demand, coupled with construction bottlenecks and unreliable project timelines, means that between 30% and 50% of large-scale data centre capacity expected to come online in 2026 could be delayed. Furthermore, the supply chain for AI infrastructure remains under severe pressure. While the focus was initially on semiconductors, by early 2026, constraints had spread to critical components such as helium, substrate materials, power management components, and thermal infrastructure. Lead times for some components now range from 20 to over 128 weeks, creating a

Canonical article URL