The Infrastructure Deficit: Why UK AI Ambitions are Stalling on Legacy Networks

The British corporate landscape is currently witnessing a paradox of digital transformation. While billions of pounds are being funneled into Artificial Intelligence (AI) initiatives, a critical component of the technological stack is being dangerously overlooked: the network.

According to a comprehensive new study from IDC, commissioned by the managed Network-as-a-Service (NaaS) provider Expereo, UK businesses are aggressively pursuing AI integration while failing to provide the robust network infrastructure required to sustain it. The report, titled “Enterprise Horizons 2026: Where Innovation Meets Reality,” serves as a stark warning to the C-suite: without a foundational shift in connectivity strategy, the promised "AI revolution" may result in little more than a series of expensive, underperforming experiments.

Main Facts: The Disconnect Between Investment and Outcome

The headline figures from the IDC InfoBrief suggest a significant gap between corporate optimism and operational reality. Despite the hype surrounding generative AI and machine learning, only 15 percent of UK organizations report that their AI implementations have exceeded expectations. Even more concerning is that a mere 3 percent claim their deployments have significantly outperformed their initial goals.

The research identifies a "reactive" investment culture as the primary culprit. Rather than building a strategic roadmap, many UK firms are throwing capital at AI in a bid to keep pace with competitors. Specifically:

• 66 percent of UK organizations are investing in AI based on its perceived potential rather than any proven return on investment (ROI).
• 15 percent of businesses admit to "aggressive spending" with almost no formal evaluation, driven largely by the fear of being left behind (FOMO).

This "gold rush" mentality has led to a situation where the software layer is advancing at breakneck speed, while the underlying hardware and connectivity layers remain stagnant. Nearly a quarter (25 percent) of businesses specifically identified inadequate network or connectivity performance as the primary reason their AI deployments failed to meet benchmarks over the past year.

Chronology: From AI Hype to the "Infrastructure Wall"

To understand how UK businesses reached this point, one must look at the trajectory of AI adoption over the last 24 months.

The 2023 Surge: The Year of Experimentation

Following the public release of advanced Large Language Models (LLMs), 2023 was characterized by a frantic rush to pilot AI programs. Boards of directors pressured IT departments to "do something with AI," often bypassing traditional procurement and infrastructure vetting processes. During this phase, network capacity was rarely a topic of conversation, as most projects were small-scale pilots or siloed experiments.

The 2024 Bottleneck: Scaling Difficulties

As organizations attempted to move these pilots into production environments in 2024, they hit what analysts call the "Infrastructure Wall." Unlike traditional cloud applications, AI workloads—particularly those involving real-time data processing and large-scale model training—demand massive bandwidth and ultra-low latency. The legacy Wide Area Networks (WANs) that served businesses well for a decade began to buckle under the strain of high-volume, bi-directional data flows.

The 2025 Outlook: A Step-Change in Demand

The IDC research indicates that we are on the precipice of an even greater strain. Approximately 55 percent of UK organizations have named AI or machine learning as their absolute top technology priority for the next 12 months. This shift suggests that network demand is no longer growing linearly; it is approaching a "step-change" where traditional incremental upgrades will no longer suffice.

Supporting Data: The Technical Barriers to Success

The IDC InfoBrief provides a granular look at why AI programs are failing to deliver. While the network is a primary culprit, it exists within a cluster of interconnected technical failures.

The Network Readiness Gap

The demand for a new type of connectivity is clear among IT leaders:

• 54 percent of UK organizations state they require more flexible and scalable networks to survive in an AI-driven economy.
• 57 percent emphasize the need for greater resilience and reliability. As AI becomes embedded in mission-critical operations (such as automated customer service or real-time supply chain logistics), any network downtime results in an immediate and total failure of the AI service.

Beyond Connectivity: Data and Cost

The report also highlighted other significant hurdles that exacerbate network issues:

1. Inadequate Data Quality (48%): AI is only as good as the data it consumes. If the network cannot efficiently transport high-quality data from the edge to the core, the AI model produces "hallucinations" or inaccurate results.
2. Unachieved ROI (45%): High operational costs, often driven by inefficient cloud egress fees and the high power consumption of AI-optimized hardware, are eating into profit margins.
3. Performance Underwhelming (42%): Many AI tools simply do not perform as advertised when integrated into a complex, legacy IT environment.

Security and Governance Concerns

As AI becomes more pervasive, the surface area for cyberattacks expands. 56 percent of UK tech leaders cited the creation of new security risks as a major future threat. Furthermore, 33 percent of respondents expressed concern about losing track of AI-related costs and ROI once the technology becomes "invisible" or embedded across various business units.

Official Responses: A Call to the Boardroom

Ben Elms, the CEO of Expereo, has been vocal about the implications of this research. According to Elms, the conversation around AI needs to move away from the "magic" of the algorithm and toward the "mechanics" of the delivery.

“Every UK organization we speak to is investing in AI, yet the data shows a clear gap opening up between AI ambition and AI outcomes,” Elms stated in response to the findings. “More often than not, that gap comes down to the network underneath. AI only delivers on its promise when the infrastructure carrying it is built to support it.”

Elms argues that the "Network-as-a-Service" (NaaS) model is no longer a luxury but a necessity for the AI era. He emphasizes that infrastructure is no longer just a "back-office" IT concern but a strategic pillar that determines the success or failure of the entire corporate strategy.

“Without resilient, scalable, cloud-optimized networks, even the most well-funded AI programmes will struggle to deliver ROI,” Elms continued. “Getting the network right is no longer an IT decision; it is one of the most important conversations happening in the boardroom today to help fulfill AI ambition.”

Industry analysts at IDC echo this sentiment, suggesting that the "Horizon 2026" timeframe will be defined by those who successfully modernized their underlying architecture and those who were strangled by legacy debt.

Implications: The Path Forward for Businesses and the Channel

The findings of the IDC/Expereo report have profound implications for the UK business ecosystem, particularly for the "Channel"—the network of managed service providers (MSPs), resellers, and consultants who advise enterprises.

1. The Strategic Opportunity for Channel Partners

For IT channel partners, the research represents a massive opening. As businesses realize that their AI investments are stalling due to connectivity issues, they will look for partners who can diagnose the "underlying connectivity story." Partners who lead with infrastructure-readiness assessments, rather than just selling AI software, are likely to become the most trusted advisors in the next two years.

2. The Shift to Network-as-a-Service (NaaS)

The data suggests a move away from static, hardware-heavy networking toward software-defined, flexible models. Because AI workloads are often "bursty"—requiring massive bandwidth for short periods—businesses need networks that can scale up and down on demand. NaaS allows companies to treat their network as a utility, aligning costs directly with usage and AI demand.

3. The Necessity of "AI-Ready" Governance

Governance can no longer be an afterthought. The research shows that 33 percent of leaders are already worried about losing control of costs. To mitigate this, companies must implement "AI-Ready" governance frameworks that monitor not just the ethical use of AI, but its technical performance and infrastructural impact. This includes monitoring data egress costs and ensuring that network security protocols (such as SASE – Secure Access Service Edge) are robust enough to handle AI-driven traffic.

4. Narrowing the Window of Opportunity

The most critical takeaway from the report is the timeline. With 55 percent of UK firms doubling down on AI in the next 12 months, the "window of opportunity" to fix the network is closing. Organizations that fail to address their infrastructure deficit in 2025 will likely find themselves with "stranded assets"—AI software that is too powerful for their networks to handle and too expensive to justify.

Conclusion: Building the Foundations for Innovation

The UK stands at a crossroads. The ambition to become a "global AI superpower" is evident in the investment levels seen across the private sector. However, the IDC and Expereo research serves as a sobering reality check. Innovation cannot exist in a vacuum; it requires a physical and digital foundation.

The "Enterprise Horizons 2026" report makes it clear that the next phase of the digital revolution will not be won by the company with the best algorithm, but by the company with the best delivery system. As AI workloads continue to scale, the network will increasingly become the differentiator between those who realize the value of their data and those who remain trapped in a cycle of reactive spending and underperforming technology. For the UK boardroom, the message is simple: if you want to win with AI, you must first master the network.