AI investment is booming, but your Enterprise Horizons 2026 research suggests many organisations aren’t seeing the returns they expected. What’s going wrong?
The conversation around AI has become dominated by what’s possible, rather than what’s practical.
Most organisations understand the opportunity. They’re investing in new models, applications and use cases because they recognise AI’s potential to transform their business. The problem is that many are scaling ambition faster than capability, without fully understanding whether their networks can support the demands AI places on data and users.
Many organisations have yet to build the foundations needed to deliver consistent returns from AI investment, with challenges around data quality, skills, governance, infrastructure and visibility often holding back ROI rather than the technology itself. As AI workloads become more distributed, organisations need a clear understanding of network performance and potential bottlenecks.
AI success starts with resilient connectivity, end-to-end network visibility and the ability to deliver consistent performance at scale.
Why do many organisations underestimate the role that network infrastructure plays in AI success?
Networks have traditionally been viewed as operational infrastructure rather than a strategic business capability. That mindset doesn’t work any more.
AI is fundamentally changing how organisations think about data. Workloads are increasingly distributed, real-time and data-intensive, so network performance becomes a direct enabler of business outcomes.
Our research found that inadequate networks are among the leading reasons AI failed to deliver expected returns. Conversely, the organisations exceeding expectations consistently cited strong network performance as one of the biggest contributors to success. That tells us that infrastructure now determines whether AI investments succeed or not.
How are AI workloads changing the demands placed on enterprise networks?
Organisations are increasingly moving large volumes of data between cloud environments, edge locations and distributed teams while expecting AI services to respond almost instantly. That places far greater emphasis on bandwidth, resilience and flexibility, which many existing networks aren’t designed to deliver.
The companies using AI in the most advanced ways already have significantly more capable networks. They understand that preparing for AI success depends on building connectivity structures that are agile, resilient and able to adapt as business needs evolve.
What separates organisations that are successfully scaling AI from those that are struggling?
The biggest difference I see lies in executional discipline. Successful organisations don’t treat AI as a collection of isolated experiments. They invest across the foundations – strengthening data quality, building internal skills, establishing governance and ensuring networks are capable of supporting AI at scale.
Our research reinforces this. The firms where AI ROI exceeded expectations were significantly more likely to have strong in-house expertise, high-quality data and robust network performance. Whereas those struggling are far more likely to cite poor data quality, skills shortages and infrastructure limitations. The winners aren’t necessarily investing more in AI. They’re preparing better for it.
Why is end-to-end network visibility becoming such a critical capability in the age of AI?
As AI workloads span multiple cloud environments, edge locations and third-party providers, organisations need far greater visibility into how applications and data are performing across their networks. Without that insight, it’s next to impossible to identify bottlenecks, resolve performance issues or understand where resilience can improve.
Better visibility allows organisations to shift from reacting to problems to proactively strengthening performance before potential issues impact AI initiatives. That’s essential when AI applications might be supporting customer experiences or business-critical decisions where downtime has immediate commercial consequences.
If business leaders are planning their next wave of AI investment, what should they prioritise to maximise long-term value?
You need to think beyond AI itself. The businesses creating sustainable value won’t necessarily be those deploying the largest models or launching the greatest number of pilots. They’ll be the ones investing in the capabilities that allow AI to scale safely and effectively over time.
My biggest piece of advice would be to ensure you have the right team, skills and governance structures in place – backed up by a resilient, secure and flexible network infrastructure.
AI investment is booming, but your Enterprise Horizons 2026 research suggests many organisations aren't seeing the returns they expected. What's going wrong?
The conversation around AI has become dominated by what's possible, rather than what's practical.
Most organisations understand the opportunity. They're investing in new models, applications and use cases because they recognise AI's potential to transform their business. The problem is that many are scaling ambition faster than capability, without fully understanding whether their networks can support the demands AI places on data and users.
