Sovereign AI Part 2 - The British AI industry

Perspective

This article explores the current position of the British AI industry and the strategic questions surrounding investment, skills, ownership and sovereign capability. It is intended to stimulate discussion and does not represent a description of services offered by Methods or formal policy advice.

Introduction

The current AI market in the UK is promising, with nearly 6,000 companies contributing £12 billion to the economy and employing over 80,000 people. The valuation of the UK’s AI market itself is £72bn, the world’s third-largest AI start-up ecosystem after the US and China. AI has the potential to transform not only the nature of work, but wider society as well, by improving productivity, enhancing our quality of living and spread prosperity more widely. The effects, if managed capably, will be felt across the financial and professional services, manufacturing, healthcare, the creative arts and academia, amongst others.

Currently, major innovations such as self-driving cars, advancements in radiography and more minute developments such as predictive text all require or rely on some form of AI. For the first, it is computer vision and object detection, for the second, it is pattern recognition machine learning models and for the last, it is Natural Language Processing. The British AI industry retains expertise in all these areas and forms of AI. As previously noted in the first article of this series, there is a strong precedent for high end technology in British history, with companies like ARM, the semiconductor company which helped pioneer modern chip designs. The government recognises the importance of this in its UK Semiconductor Strategy report of 2023.

What is required to sustain the aforementioned growth of the industry is major capital investment, the training of an AI and data workforce, and a focus on maintaining and expanding British company ownership in this field.

Capital investment:

In April 2026, the government launched Sovereign AI, a £500 million initiative designed to support the growth of British AI companies through a combination of investment, talent and access to advanced computing resources. While this represents a significant commitment, the scale of investment needs to be in the billions to reflect the growing costs of compute, the range of companies requiring support and the depth of technical research taking place within British universities.

The AI superpowers of the world, China and the United States, are investing at enormous scale in AI infrastructure and development. Reuters reported that Alphabet, Amazon, Meta and Microsoft alone were expected to invest around US$650 billion in AI-related infrastructure during 2026. While the UK cannot compare in absolute terms, it can increase its spending relative to GDP, reflecting the reality that AI is the new frontier technology of the 21st century.

Sovereign AI

Sovereign AI’s purpose is to supercharge the growing AI industry in the UK and enable British businesses to attract more talent, acquire funding, and have access to the UK’s supercomputers. The Government’s Department for Science, Innovation and Technology (DSIT) says it will be independent, with an investment committee making decisions as to who to fund and why, without political interference. The fund has already begun investing in Callosum and Ineffable, two British AI companies, while also developing compute agreements with others like Prima Mente, Doubleword, Cosine, Cursive, Odyssey and Twig Bio. While this is a start, more is required.

Moreover, public-private partnerships (PPPs) need to be established with the capital and access to supercomputers required, but with greater alignment with the state’s strategic priorities. In this sphere, complete independence may not necessarily be a virtue, as it can lead to misalignment. Establishing PPPs, which have the explicit goal of developing Sovereign AI by the middle of the century, should be at the forefront of strategic thinking.

AI Trained Workforce

The UK, like many other countries, still has significant work to do in preparing its workforce to use AI effectively and understand its impacts and risks. For instance, the Nash Squared Digital Leadership Report found that 59% of organisations in the UK are not upskilling employees in generative AI at all, which is quite high given generative AI is one of the primary forms of AI used both by the general public but also workers across various industries due to its relatively low cost of entry and usefulness for automating previously manual writing and research tasks.

Moreover, Skills England reported that an estimated 71% of the country’s workforce in 2035 are those people who are already in the workforce. Therefore, those requiring upskilling are people with established skills and careers, not simply new entrants. Initiatives therefore must be directed at those who may not prioritise their use of AI or may use it in ad hoc manner only. Training a workforce in AI also means making people aware of data security and privacy risks. Information entered into some AI systems may be retained, processed or used in ways that are not immediately obvious to users, making it important to understand how different tools handle data. A high profile example of this was the sensitive information leaked to OpenAI by Samsung workers while they used ChatGPT to work through engineering problems.

British ownership

Maintaining ownership of companies which are working on AI research is another crucial step. This maintains control over the IP as well as any hardware developed and recognises AI as a crucial element of both the economy and national security. This prioritisation helps signal the importance of the growing field to both government agencies involved in supporting AI development as well as private investors.

DeepMind, for instance, which pioneered the development of AlphaGo and AlphaZero, was acquired by Google in 2014 and was then merged with Google Brain. ARM, with its innovative chip designs which served as a model for other companies, was acquired by Japanese company Softbank. The loss of these companies affects British industry, and retaining control of such crucial companies should be incentivised using subsidies and tax breaks. This includes incentives for hiring and training British graduates, as well as offering schemes for foreign workers to settle and reside in the UK.

Finally, the development of a sovereign frontier AI model is crucial. The latest encouraging sign on this front comes from British start-up Cosine, which will work with Babcock International Group, BT, Lloyds Banking Group, London Stock Exchange Group (LSEG), NatWest Group, PwC, Thales UK, and Telefónica Tech UK&I on designing a sovereign frontier model by the name of Lumen. According to Cosine, Lumen would be Britain’s first sovereign frontier AI model. Cosine is one of the companies supported through the government’s aforementioned Sovereign AI initiative. The model will be trained on the supercomputer Isambard-AI. This is a significant step forward and it is worth watching its development in the near future.

The British AI industry is one of the best established in the world. It continues to grow and has immense potential for further development. There are several routes forward which would ensure this remains the case, including retaining British ownership of AI companies, pushing frontier AI models, training our workforce and making certain the right capital investment exists.

About this article

This article presents the author’s perspective and is intended to encourage discussion about the future of the British AI industry and sovereign capability. It should not be interpreted as government policy advice or as a statement of Methods’ service capability.