Nearly half (47 per cent) of UK businesses lack the network infrastructure to support new technology initiatives, including AI, according to new research.
Meanwhile, 49 per cent of organisations reported that insufficient network performance prevented or limited their capacity to support large data and AI projects.
The findings were revealed in Expereo’s Enterprise Horizons 2025 study of 650 global enterprise technology leaders across Europe, the US and APAC, showcasing the roadblocks to UK AI plans.
The study found that two-fifths of UK CIOs have warned of unrealistic expectations of AI and how it will impact business performance. 26 per cent of tech executives believed that these AI expectations are growing faster than their ability to meet them, despite receiving greater recognition for successful projects.
Richard Petrie, LINX CTO, commented:”These findings reveals that the UK’s network infrastructure is struggling to meet the growing demands of AI, with nearly half of UK organisations reporting network performance issues that hinder large AI projects. To support the UK’s goal of becoming a global AI hub, enhancing infrastructure is critical. It is essential to invest in high-capacity, secure, and energy-efficient networks in order to help businesses overcome these challenges. With the government’s push to increase infrastructure capacity, LINX is well-positioned to lead the way in building the robust digital infrastructure necessary to support AI at scale and maintain the UK’s competitive edge.”
Business leaders are undoubtedly seeing the benefits, with 88 per cent labelling AI as important to fulfilling business priorities over the next 12 months. Customer-facing activities (64 per cent) and cost savings (65 per cent) were cited as the areas where AI is already having the most positive impact on businesses.
Sachin Agrawal, Managing Director for Zoho UK, commented: “As businesses in the UK and the rest of the world rush to embrace AI, the reality check from the research is clear that without strong, scalable infrastructure, the promise of AI could become a liability rather than a strength. Network performance forms part of the bedrock for AI success, especially when it comes to handling large-scale data processing and real-time applications.”
“Firm consideration should be taken to right size AI models. Different language models excel in different use cases. Not all AI use cases demand the use of LLMs, which require the most power and cost and therefore place the most pressure on infrastructure. In fact, narrow, small or medium language models can be fine-tuned to deliver more domain-specific business benefits and can reduce dependency on LLMs.
“As AI adoption accelerates, global collaboration will be required by industry experts, government regulators and academia and these entities should work together to establish frameworks that promote safe and ethical use of AI systems. Without clear standards for transparency, fairness, and accountability, organisations risk undermining trust, both internally among employees and externally with customers and regulators.”
“We’re seeing platforms like Zoho Creator demonstrate how AI can be democratised and made more actionable, safely and securely. By enabling faster, smarter app development without extra cost or complexity, tools like these help bridge the gap between ambition and execution. As AI adoption continues to scale, strategic investments in both infrastructure and intelligent development platforms will contribute to long-term success.”
Despite the opportunities ahead, 14 per cent of UK business leaders believed that AI had fallen short of expectations so far.
41 per cent of businesses highlighted concerns around AI governance, with ethics a significant obstacle to AI initiatives. These issues are compounded by staff resistance to AI adoption and concerns of external technology partners lacking the right capabilities to support AI implementation.
Arkadiy Ukolov, Co-Founder and CEO of Ulla Technology Ltd, commented: “These findings highlight the significant challenges UK businesses face with AI adoption, particularly around network infrastructure and data security. The fact that nearly half of UK organisations report that their current network infrastructure is not ready to support AI initiatives, while over 40% express concerns about AI governance and ethics is alarming.”
“These issues emphasise the urgent need for solutions that address both security and infrastructure limitations. By processing data locally, businesses can gain greater control over sensitive information, directly addressing the governance and ethical concerns raised by many organisations. Additionally, this approach helps reduce the strain on network infrastructure, offering a practical solution for organisations struggling to support AI workloads.”
