Starling is building a transactional foundation model (TFM) with support from NVIDIA that will be the first banking model to be trained and hosted within a UK sovereign cloud.* Working with UK cloud provider Civo, Starling is supercharging its ability to design tools for customers while also supporting the growth of UK cloud capability.

The model will be trained on proprietary Starling data from over 15 billion anonymised transactions to give the bank deeper insight into the needs and behaviours of its customers. In time, it will form a single intelligence layer – ‘Starling Intelligence’ – that elevates the bank’s AI capabilities and enables it to build entirely new experiences.

“We’ve made huge strides with our proprietary AI tools, launching the UK’s first agentic AI financial assistant and rolling out new features every week. Now we want to connect these tools with a single intelligence layer that’s trained entirely on the specific needs of our customers,” said Harriet Rees, Group CIO at Starling and UK Government AI Champion for Financial Services. “We also want to take a leadership role in the development of home-grown cloud infrastructure, which is essential to European companies’ ability to diversify their suppliers and reduce concentration risk.”

By training its TFM on Civo’s UK cloud, Starling is following recommendations made in the UK Government’s Financial Services AI Adoption Plan, co-authored by Harriet Rees and published in July 2026. The plan flags a reliance on a small number of global AI and cloud providers and recommends that government and industry broaden their partnerships to improve optionality and systemic resilience.

“AI will only deliver lasting value for the UK economy if it’s built on secure, resilient, sovereign infrastructure,” said Mark Boost, CEO of Civo. “Starling is leading the financial sector by training and hosting its Transactional Foundation Model on a homegrown cloud. Pairing NVIDIA compute with Civo’s UK sovereign cloud shows that banks don’t have to trade performance or innovation for control of their data, or depend on a handful of global providers.”

The TFM will be developed with NVIDIA’s financial services team, using NVIDIA’s Transaction Foundation Model developer blueprint, including NVIDIA CUDA-X libraries for GPU-accelerated data processing and NVIDIA NeMo AutoModel for model training.

Optimised for high-performance AI workloads, Civo’s sovereign infrastructure, from NVIDIA Hopper GPUs to upcoming Vera Rubin deployments, provides the low-latency compute and advanced liquid cooling essential for real-time reasoning, autonomous systems, and agentic AI. It will create a reusable intelligence layer capable of identifying patterns in anonymised transaction data and supporting multiple applications.

“Banks are at the forefront of turning massive amounts of data into timely, personalised insights while also protecting customers from fraud,” said Aser Blanco, Global Head of Banking at NVIDIA. “By combining NVIDIA’s accelerated computing and software for data processing and model training with Civo’s UK sovereign cloud, Starling can deliver exceptional customer experiences while maintaining control of its customers’ data.”

Starling intends to put its TFM into production in early 2027, initially to improve existing AI services around fraud detection and then to augment Starling Assistant, the UK’s first agentic AI financial assistant. The Starling Intelligence layer will enable Starling Assistant to provide more personalised experiences, such as proactively serving bespoke budgeting guidance.

*Starling is the first bank to train a TFM on UK cloud infrastructure owned by a UK-owned and headquartered company, meaning all data is within UK jurisdiction only