Training large language models grabbed most of the AI infrastructure conversation for the past three years. The next phase is different. As organisations shift from building AI to running it at scale, the infrastructure question is moving from centralised data centres to where inference actually happens — and colocation is back in the frame.
Inference AI — running a model to generate real-time outputs and decisions — depends on latency in a way training does not. Fraud detection, computer vision, industrial automation, and AI-assisted decision-making all fail usability thresholds if response times slip. That constraint is pulling AI workloads toward regional infrastructure rather than the nearest hyperscale availability zone.
The data privacy picture is adding pressure. Sending proprietary datasets and sensitive prompts into shared hyperscale environments is increasingly a board-level concern, not just an IT configuration question. As AI systems process competitive data and intellectual property, sovereignty questions are reaching executives who delegated cloud decisions entirely five years ago.
Research from Vanson Bourne found that 87% of enterprise CIOs plan to move away from public cloud — partially or fully — over the next two years. Gartner projects 20% of existing workloads will migrate from global public clouds to local or regional alternatives by the end of 2026.
Mark Lewis, Chief Marketing Officer at Pulsant, argues that the economics reinforce the case. AI inference workloads running continuously in public cloud create operational expenditure uncertainty that training-cost discussions rarely surface. Colocation offers predictable costs and greater control over network configuration — particularly relevant for organisations running latency-sensitive inference at scale.
The pattern emerging across the UK and Europe is less about rejecting public cloud than about finding where each workload fits best: hyperscale for training and storage, regional colocation for inference where latency and data residency matter.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
