As the US and China pour hundreds of billions into artificial intelligence, a quieter revolution is taking shape in Europe — one built not on brute computing force but on smart, specialised tools. From a legal AI platform valued at $8 billion to a European-built answer engine rewriting how we discover information, the era of vertical AI is here. Britain’s place in it is anything but guaranteed.
The question used to be whether artificial intelligence would eventually come for your job. That debate is largely over. The question now is how quickly, in which sectors, and whether the country you live in will be positioned to catch you when it does – or create something better on the other side.
In the United Kingdom, the answer is complicated. On paper, Britain’s AI sector looks impressive. The government’s own 2024 AI Sector Study estimates the market at $92 billion—larger than in any other European country — and reports a record £2.9 billion in private investment in a single year, surpassing even the tech boom highs of 2022.
Over 5,800 AI companies now operate across the country, employing more than 86,000 people, a workforce that has grown by 72% since 2022. By some measures, London remains, as one investor put it in the report, “still the number one city” for AI in Europe, ahead of Berlin, Paris, and Stockholm.
But scratch beneath those headline figures and a more ambiguous picture emerges. Britain’s private AI investment in 2024 — approximately $4.5 billion — is roughly 24 times smaller than America’s $109 billion, according to Stanford University’s 2025 AI Index Report. The gap in generative AI specifically is even more stark: US spending exceeded the combined total of China, the EU, and the UK by over $25 billion. The race, in other words, is not close. And for workers whose roles sit squarely in AI’s crosshairs, the question of whether Europe and Britain can build a competitive, sovereign AI capability is not abstract — it is intensely personal.
“We are at a sliding doors moment.” – IPPR, on AI’s impact on the UK labour market
The Jobs at Stake
The Institute for Public Policy Research (IPPR) has modelled a range of outcomes for what generative AI will mean for British workers, and none of them are entirely comfortable reading. In its central scenario, around 545,000 jobs could be lost — but GDP could rise by 3.1%, equivalent to £64 billion per year. In the worst case, 1.5 million jobs disappear with no corresponding economic gain. In the best case, no net jobs are lost, and GDP rises by 4%, or £92 billion annually. The critical difference between those outcomes is policy.
The Tony Blair Institute for Global Change puts the potential displacement figure even higher, estimating that between one and three million UK jobs could ultimately be affected over the coming decades, with peak losses running at 60,000 to 275,000 roles per year. Administrative staff, customer service workers, data analysts, and professionals in banking and finance face the sharpest exposure. McKinsey’s analysis of the UK labour market, published in mid-2025, found that job advertisements for roles with high exposure to AI and large language models had already fallen by 38% since 2022 — compared to a 21% drop for roles with low AI exposure. Something is already happening. It is just happening unevenly.
The World Economic Forum’s Future of Jobs Report 2025 offers a more optimistic global frame: 92 million jobs may be displaced by 2030, but 170 million new roles could be created — a net gain of 78 million. AI development, cybersecurity, and sustainability are projected to be the fastest-growing job categories. Whether Britain captures those new roles, however, depends entirely on what its companies build — and how ambitiously.
The Vertical AI Revolution
Lost in the conversation about foundation models – OpenAI, Google DeepMind, Anthropic – is a quieter but arguably more commercially important story: the rise of vertical AI. These are not general-purpose chatbots. They are narrowly focused, domain-specific tools that answer questions, surface information, and complete professional tasks within a defined field. And they are attracting extraordinary investor interest.
The clearest example is Harvey, a San Francisco-based legal AI platform that has rapidly become one of the most-watched companies in enterprise technology. Founded in 2022 by Winston Weinberg, a former litigator at O’Melveny & Myers, and Gabriel Pereyra, a research scientist who previously worked at DeepMind and Meta, Harvey was built on a deceptively simple insight: the legal industry is almost entirely made of words. Research, drafting, contract review, due diligence – all of it is language. And large language models are extraordinarily good at language.
The market agreed. Harvey raised $300 million in February 2025 at a $3 billion valuation, followed by another $300 million just four months later at $5 billion, and then a further $160 million round led by Andreessen Horowitz that valued the company at $8 billion by the end of the year. Total capital raised has exceeded $1.2 billion.
By August 2025, Harvey had surpassed $100 million in annual recurring revenue and counted 50 of the AmLaw 100 firms – the most prestigious law firms in the United States — among its customers, along with corporate legal departments at companies including PwC and KKR. Its technology is now used by approximately 100,000 lawyers across 60 countries.
Harvey’s rise from zero to $8 billion in three years is a masterclass in what vertical AI can achieve when domain expertise meets frontier models.
Harvey’s ascent illustrates a broader principle: the most durable value in AI will not necessarily come from building the most powerful model. It will come from knowing an industry deeply enough to deploy AI in ways that practitioners actually trust. The platform’s approach reflects a maturing understanding of how enterprise AI actually works in practice.
Europe’s Answer Engine: What marvn.ai Is Building
While Harvey has attracted the lion’s share of headlines, Europe is generating its own vertical AI story – one that demonstrates how the ‘answer engine’ model can be applied to information-dense, high-stakes environments beyond law.
marvn.ai represents a distinct approach to the vertical search and discovery problem. Rather than returning a list of links in the manner of a traditional search engine, marvn functions as an AI-powered answer engine: users ask a question, and the system draws on a curated, proprietary database to generate a direct response, surfacing relevant information and enabling follow-up dialogue. The platform has built partnerships with over 500 brands and continues to expand its underlying database infrastructure in what the company describes as an ongoing investment in technical talent and data quality.
In January 2026, marvn launched a ‘Discover‘ section, a news and knowledge hub that searches reputable sources, generates contextual overviews, and allows users to ask follow-up questions on what they read. From a product standpoint, this positions marvn not merely as a search utility but as a discovery and research companion: a tool that creates reasons for users to return even when they are not executing a specific query.
The economic logic is sound. Higher session frequency reduces customer acquisition costs over time. Better-matched users generate higher-quality outcomes for platform partners. And an answer engine that users trust tends to become embedded in their decision-making process in ways that traditional search never does.
The underlying proposition of marvn’s vertical search model aligns with a principle that investors across the AI landscape are increasingly recognising: proprietary data is the real moat. Anyone can bolt a large language model onto a public dataset. The differentiation comes from curated, validated, domain-specific information that improves the quality and reliability of AI-generated answers. marvn’s investment in its proprietary database is, in this sense, an investment in competitive durability.
The most valuable AI companies of the next decade will not be those with the biggest models. They will be those with the best data — and the sharpest domain focus.
What marvn.ai demonstrates, alongside Harvey in legal services, is that European builders are not simply watching the AI revolution unfold from the sidelines. They are constructing specialised tools for specific information environments, and they are doing so with a clear-eyed understanding of the unit economics that differentiate a sustainable AI business from a well-funded experiment.
The European Context: Investment, Regulation, and the Sovereignty Question
Europe’s aggregate AI investment picture is more complex than its ambition. In February 2025, the EU launched InvestAI, a plan to mobilise €200 billion in AI investment and establish a new €20 billion fund focused on AI gigafactories. It is a substantial commitment. But the starting position matters: in 2024, AI funding across Europe – including the UK – still represented only 25% of global venture capital flows into the sector, compared to 42% for the United States alone.
Within Europe, the UK’s position is also shifting. In 2024, the UK attracted 51 AI-related inward investment projects worth more than £15 billion, with major announcements from Amazon, Google, CoreWeave, and Vantage Data Centres. AI-related business revenues in Britain more than doubled between 2022 and 2024, from £10.6 billion to £23.9 billion. The number of AI companies grew by 85% over the same period.
Yet the country faces structural headwinds. A fifth of all AI inward investment since 2019 arrived in 2024 alone but the report also notes that access to scale-up and late-stage capital remains a persistent barrier for UK-based AI firms trying to expand globally. A significant proportion of UK AI infrastructure remains hosted on American cloud platforms. ‘We have some good colocation data centres,’ one investor in the government study conceded, ‘but most of our data is stored with AWS or Google.’
The regulatory divergence between the UK and EU also creates uncertainty. Britain’s current approach offers flexibility to innovate but risks creating complexity for companies operating across both markets. What the UK AI Act will ultimately look like, and whether it will align more closely with Brussels or Washington, remains an open question with significant implications for the sector’s architecture.
The Bigger Picture: Winners, Losers, and What Comes Next
For workers, the analytical consensus is clearer than the political debate: AI’s impact will be uneven, sector-specific, and heavily dependent on which organisations they work for. Research from the Centre for Economic Policy Research (CEPR) found that in the short term, AI is acting as a complementary input without immediately reducing employment. But the authors were careful to note that this capital-deepening effect may be transitional: as systems become more capable and companies deepen their integration, labour-displacing effects are likely to emerge.
Goldman Sachs estimates a 0.5% rise in unemployment during the AI transition period — modest by historical standards, and consistent with past technology adoption cycles. But past cycles did not move this fast, and they did not target the knowledge-economy jobs that form the backbone of Britain’s post-industrial workforce. Legal secretaries, data analysts, financial services professionals, junior researchers, and entry-level graduates in media and design are all in sectors where AI exposure is high, and job advertisements are already contracting.
The optimistic reading is that AI creates as many jobs as it destroys, and the WEF’s projections support that view at a global level. The more difficult question is whether those new roles appear in the same countries, the same cities, and among the same workers as the ones displaced. History suggests the answer is often no, at least not immediately. The challenge for Britain, and for Europe, is to build the policy frameworks, the education pipelines, and the homegrown AI industries that tilt those probabilities in their citizens’ favour.
Vertical AI companies like Harvey and marvn.ai are not just commercial bets. They are, in a broader sense, proof of concept: evidence that focused, domain-specific AI built on proprietary data can compete at the highest level, generate sustainable revenues, and create new categories of highly skilled employment. The question for policymakers and investors in Britain and across Europe is whether they will back enough of them to matter.
The Bottom Line
Britain is not losing the AI race. But it is not clearly winning it either. The headline numbers are strong. The talent is here. The venture ecosystem is growing. And European founders are building serious companies — from legal AI infrastructure valued in the billions to vertically focused answer engines that are rewriting how millions of people find and process information.
What the UK cannot afford is complacency. The gap between American AI investment and European AI investment is not closing – it is widening. The window to establish European sovereign capability in the systems that will define the next economy is real, but it is not unlimited.
