British companies continue to pour money into AI, even though most are failing to achieve significant AI productivity gains at scale, according to new research published today by Snowflake.
The survey, conducted by YouGov on behalf of Snowflake and surveyed 500 UK business decision makers, found that only 23% of organizations have achieved large-scale AI-driven productivity improvements, with a further 45% saying improvements are still limited to specific or experimental use cases.
Despite this, the appetite for spending on AI shows no signs of slowing down. The survey found that only 1% of organizations plan to reduce their AI investments over the next 12 to 24 months, suggesting that confidence in AI’s long-term potential remains strong even if short-term results prove difficult to achieve.
This finding applies at a critical policy moment focused on AI as an economic tool. The UK government’s AI Opportunities Action Plan aims to generate economic growth of £47bn a year and estimates that the uptake of AI could increase national productivity by up to 1.5% each year. For organizations that have not yet taken action, the possibilities may be narrowing.
To find out more about why UK organizations are struggling to turn their AI investments into tangible results, read our analysis of the key AI and automation trends shaping 2026.
Internal barriers, not technology, are slowing AI productivity across the UK
This report challenges the common assumption that technology readiness is holding organizations back. Only 19% of respondents cited technology as a barrier to progress. Rather, the main obstacles are skills shortages, poor data quality, organizational silos, and unclear strategic direction.
Governance also emerged as a structural weakness. Only 24% of organizations say they prioritize their AI initiatives using a rigorous framework aligned to business goals, meaning the majority of adoptions lack a clear strategic rationale. Responsibility for AI governance is typically fragmented among executives, technology, data, and business leaders, with no single clear owner, slowing decision-making and limiting accountability.
This pattern is consistent with extensive research into the impact of AI on the UK workplace, which finds that organizations that invest in AI without a strong governance framework often see their efforts strengthened, not diminished.
Dr Fabian Stefani, an economist and undergraduate research lecturer at the Oxford Internet Institute at the University of Oxford, said the findings were consistent with historical patterns around transformative technologies. He commented:
“Technology advances rarely translate into immediate productivity gains because organizations take time to adapt their workflows, governance structures, and capabilities.”
Dr Stefani also pointed to the SkillScale Research Group’s findings that workers with AI-related skills already command a wage premium of around 23% in the UK, pointing to skills as an important and growing constraint, as well as improved employment prospects and additional benefits. For organizations that are slow to build these capabilities, the talent gap (and productivity gap) is likely to widen further.
“Expanding access to AI skills and training will be critical if organizations want to maintain and scale productivity gains.”
Which UK industries are winning and losing in the AI productivity race?
The research highlights significant differences in AI maturity across industries in the UK. Financial services organizations are more advanced when it comes to governance and strategic alignment, but regulatory and reputational concerns have slowed their move to scale. Manufacturing companies express strong confidence in AI’s long-term potential, but expect returns to slow due to skills gaps and integration challenges. Retail lags in both reliability and delivery, and AI is often limited to isolated use cases amid persistent data quality issues and fragmented ownership.
The public sector presents perhaps the most cautious picture. Approximately 52% of public sector leaders say AI will not materially improve productivity for at least two years, 66% report that ethics and safety concerns will heavily influence their adoption decision, and 53% cite reliability of AI output as their top concern. While this prudence reflects a responsible approach to risk, there is also a risk that significant efficiency gains will be left unrealized as other sectors accelerate. This tension is already visible in broader workplace analysis data for 2026.
Cutting costs over growth: How UK businesses measure AI success
When measuring the value of AI, cost savings lead the way. Almost half (44%) of respondents cited it as their most important measure of success, above 26% revenue growth. Approximately 40% of all organizations surveyed expect it to take more than two years for AI to realize material productivity gains.
These findings align with UC Today’s own analysis of how UK organizations are evaluating AI platforms in 2026. The analysis found that buyers are increasingly demanding proof of operational benefits rather than accepting vendor promises at face value.
Jennifer Bellicent, principal data strategist at Snowflake, said the research shows a clear gap between ambition and execution. She said:
“Improving productivity requires clear ownership, a strong data foundation, and alignment between AI initiatives and measurable business objectives. The focus must now shift from experimentation to disciplined execution.”
For UK organizations yet to take advantage of AI, the message from Snowflake’s research is that the technology is ready. The question is whether it is.
