Microsoft has withdrawn from an agreement to become the anchor customer for a planned AI data center in Loughton, Essex, after delays in securing an electricity connection pushed the site’s expected power availability into the 2030s. The decision highlights how access to grid power has become a decisive constraint on AI infrastructure expansion.telegraph.co+1

A high-profile UK project

The Loughton facility was announced in 2025 during President Donald Trump’s state visit to the United Kingdom. It was intended to provide 50 megawatts of AI computing capacity, with potential expansion to 90 megawatts using 23,040 Nvidia GB300 processors.telegraph.co

The project was originally expected to begin operating in early 2027. Microsoft was set to be the main tenant, using the capacity to support Azure AI workloads in the UK.finance.yahoo

Nscale, a British AI infrastructure company, owns and is building the data center. The company has positioned itself as a UK-based provider of sovereign AI infrastructure and has received backing from Nvidia.telegraph.co

Why Microsoft walked away

The central issue is electricity. Grid officials reportedly told Nscale that the Loughton site would not receive sufficient power until the 2030s, several years beyond the facility’s original deployment timeline.telegraph.co+1

For Microsoft, that delay made the site unsuitable for near-term Azure capacity planning. Large-scale AI deployments require not only GPUs, networking, and cooling, but also reliable access to substantial electrical power—often for years in advance.finance.yahoo

Microsoft said its decision related only to the Loughton site and emphasized that its broader partnership with Nscale remains intact.telegraph.co

What this means for Nscale

The withdrawal is a setback for Nscale, which had counted on Microsoft as the only announced customer for the Loughton facility. Microsoft had agreed to rent capacity based on roughly 23,000 Nvidia AI chips at the site.telegraph.co

Nscale says its relationship with Microsoft remains strong and that the London-area grid issue does not affect its other contracts. The company says it remains committed to the Loughton project and is exploring alternative power sources, as well as additional expansion sites in the UK.telegraph.co

The timing is challenging because Nscale recently filed for a U.S. initial public offering. Microsoft’s exit could complicate investor assessments of the company’s ability to deliver planned infrastructure on schedule, particularly given its rapid growth and reliance on a small number of major customers.telegraph.co

Nscale’s total contracted value has grown from $100 million to more than $103 billion in roughly two and a half years. Microsoft and Anthropic together account for approximately 85% of its contracts, making customer concentration an important risk factor.telegraph.co

Power is becoming the bottleneck

The Loughton case shows that AI infrastructure constraints are not limited to the United States. Access to electricity, grid connection timelines, and energy prices can be as important as access to advanced processors, capital, or construction capacity.telegraph.co

AI data centers consume far more power than traditional facilities because they rely on dense clusters of GPUs and specialized accelerators. They also require advanced cooling systems to manage the heat generated by high-performance hardware.networkworld

In the UK, high electricity prices and long waits for grid connections have already affected other AI projects. OpenAI previously suspended plans for a “Stargate” AI facility with Nscale in Tyneside, citing Britain’s high electricity costs.telegraph.co

A warning for AI buildouts

Microsoft’s retreat does not mean the company is reducing its overall AI infrastructure investment. Rather, it suggests that hyperscalers are becoming more selective about where they commit capacity, prioritizing sites with realistic power availability.finance.yahoo

For data center developers, the lesson is clear: securing land, chips, and financing is no longer enough. A project must also have a credible path to grid power on a timeline that matches customer demand.

For governments hoping to attract AI investment, the case underscores the need to accelerate grid planning and energy infrastructure. Without reliable power access, even well-funded AI data center projects can stall before they begin operating