- Microsoft’s UK South Azure region is at full capacity, blocking new VM and Kubernetes deployments for customers spending millions per year on the platform.
- Azure support staff are recommending customers move workloads to Sweden—but regulated industries can’t legally ship data offshore.
- The bottleneck stems from surging AI demand, and insiders say capacity won’t ease until October 2026 at the earliest.
Microsoft’s UK Azure regions are full. Not “getting busy” or “running tight”—full. Customers spending millions of pounds annually on Azure are being told they cannot provision new virtual machines or Kubernetes clusters in either the UK South or UK West regions, the only two Azure data center zones available in Britain.
A Register reader whose firm spends multi-million pounds on Azure annually reported that a Microsoft architect confirmed zero available capacity in both UK regions. AMD-powered instances are the hardest hit, along with high-performance computing and GPU-accelerated virtual machines—the exact workloads AI teams need most.
Microsoft’s response has been to point customers elsewhere. Support staff are recommending a migration to Sweden, a suggestion that works fine for a hobby project but falls apart the moment you’re dealing with regulated data. “Pushing their datacenters to capacity with no real plan to build out or expand them is just piss poor planning,” one affected user wrote online.
Why the UK Azure Capacity Crunch Matters
The capacity shortage isn’t just an inconvenience—it’s a sovereignty problem. Mark Boost, CEO of cloud platform Civo, told The Register: “When organizations are told to move workloads outside the UK due to capacity constraints, it stops being just a technical issue and becomes a sovereignty question.” Healthcare providers, financial institutions, and government contractors can’t simply relocate patient records or transaction data to a Swedish data center to keep their applications running.
This isn’t the first time Azure has run out of room. In 2020, the words “Azure seems to be full” echoed across forums as COVID-era remote work demand overwhelmed Microsoft’s infrastructure. The current crunch has a different culprit: AI. The insatiable demand for GPU compute and high-memory VMs—driven by model training, inference workloads, and agentic AI platforms—has consumed capacity faster than Microsoft can build it.
Microsoft is spending big to fix the gap. The company shelled out $6.7 billion on data center leases in Q2 FY2026 alone and has committed to over $50 billion in future leases. In the UK specifically, Microsoft is building a data center campus at the former Skelton Grange power station in Leeds and an AI supercomputer at an Nscale facility in Loughton, Essex. The company is also developing sites near Newport, Wales.
The Broader AI Infrastructure Bottleneck
The UK pinch is part of a global pattern. Commenters on Reddit and social media report similar constraints in UK West, North Europe, and several US regions. Data Center Dynamics reported that customers are being turned away from increasing compute across multiple availability zones, with one zone hit harder than others.
The math is brutal. AI workloads don’t just need more servers—they need specialized hardware (GPUs, high-bandwidth memory) that takes months to source and install. Microsoft’s data center buildout is already running behind schedule across the US, with nearly 40% of planned facilities delayed. The UK, with its stricter planning regulations and limited available land, faces an even tighter timeline.
Insiders close to Microsoft tell The Register that the company is working on the problem and expects capacity to ease around October 2026. Until then, UK Azure customers have three options: hope their Microsoft representative grants an exception, migrate to a non-UK region and accept the compliance risk, or start evaluating alternatives.
The UK South region was launched in 2016 with additional availability zones added in 2019. Microsoft also operates UK West, based near Wales. Both regions are now at capacity.
