Summary
- Colt's survey found 31% of respondents prioritising security and resilience, compared with 25% prioritising additional AI capabilities.
- Some 96% reported AI-related infrastructure challenges, including cybersecurity exposure, network traffic, bandwidth, and cloud connectivity.
- Although 71% expressed confidence about future readiness, the findings represent self-reported perceptions rather than independent technical assessments.
Security and infrastructure resilience have overtaken the expansion of artificial intelligence capabilities as a leading technology priority among large businesses surveyed by Colt Technology Services, as organisations confront the demands created by putting AI applications into operation.
The network and digital infrastructure provider surveyed 1,020 IT decision-makers across ten countries, finding that 31% named security and resilience as a top priority. Adding AI and machine learning capabilities was identified by 25%, a decline of nine percentage points compared with the previous year’s research.
Almost every respondent, 96%, reported infrastructure challenges associated with AI, although 71% said they were confident their organisations could meet the technology’s requirements over the next two to three years.
The findings indicate that businesses are reconsidering the infrastructure required to sustain AI adoption, rather than concentrating solely on introducing additional applications. They also reveal a difference between the confidence respondents express about their future readiness and the operational pressures they report experiencing today.
AI changes network and security requirements
Cybersecurity exposure was the most frequently reported infrastructure challenge, identified by 40% of participants. Increased network traffic followed at 39%, while 37% cited higher bandwidth requirements and 36% reported additional demands involving cloud connectivity.
Lower latency and greater automation were each identified by 35% of respondents, illustrating the range of technical requirements associated with AI workloads.
Enterprise applications frequently depend on communication between internal systems, external model providers, cloud platforms, and employee devices. Where AI processes involve repeated requests or large volumes of information, their operation can increase the demands placed on networks and supporting infrastructure.
Different applications create different requirements. An employee using an AI assistant occasionally to summarise a document may generate comparatively limited network traffic, while an automated system coordinating repeated actions across several enterprise applications could require more frequent exchanges of information.
Applications involving real-time processing can also be sensitive to latency, particularly where a delay affects the availability of a wider business service. The suitability of existing infrastructure therefore depends on workload characteristics rather than the presence of AI alone.
Adding external services also changes the systems that security teams must monitor. Information may pass through additional application interfaces and suppliers, while automated workflows can operate using permissions granted to non-human identities.
These arrangements create security and governance requirements that may not have existed when the underlying business applications were originally developed.
Investment shifts towards operational controls
Colt found that 34% of respondents were investing in cybersecurity and threat detection, while 31% identified investment in AI governance, risk, and compliance.
The figures indicate a continuing commitment to technology spending, but with greater attention to protecting and managing systems already being introduced. Some 70% of respondents expect their overall technology budgets to increase during the next three years.
That expectation does not mean businesses are planning to direct the additional expenditure exclusively towards security. Infrastructure modernisation, cloud services, AI applications, and other priorities may all contribute to future technology budgets.
Security requirements also vary according to how applications are deployed. A business relying on external AI services may need to assess supplier access, the movement of information, and the resilience of network connections, while an organisation operating models internally may face different demands involving computing capacity and infrastructure management.
Where several systems are connected through automated workflows, businesses must also establish which component is responsible when an operation fails. An interruption affecting an external service can prevent a process from completing even when the organisation’s own applications remain available.
This creates operational dependencies that may not be apparent when individual services are evaluated separately.
The consequences extend into incident response and recovery planning. Businesses need to understand which applications depend on external connections, how those dependencies are monitored, and what happens when one of the services becomes unavailable.
Confidence remains high despite reported problems
Although 96% of surveyed organisations identified AI-related infrastructure challenges, 71% expressed confidence that they were prepared for future requirements.
Respondents in the United States reported the highest confidence at 77%, followed by those in the UK and Europe at 74%. The corresponding figure in Asia was 64%, while cybersecurity exposure was identified as a challenge by 47% of participants in that region.
These findings are not necessarily contradictory. An organisation may recognise current limitations while believing that planned investments or existing technical arrangements will allow it to manage additional demand.
However, the survey does not establish whether the respondents’ organisations have tested those arrangements under realistic operating conditions. Confidence in a recovery plan is different from evidence that critical applications can continue operating during a significant infrastructure disruption.
Recent incidents provide practical context for that distinction. Techopia’s examination of a Google infrastructure outage in London considered the difficulties that can emerge when technical recovery arrangements do not behave as expected.
AI applications may introduce additional dependencies on model providers, external databases, network connections, and cloud services. Each of those components can affect the reliability of a process, even where the individual applications have been designed with resilience in mind.
The need to coordinate them increases as businesses introduce automation spanning several departments or technology environments.
Regional priorities reflect different approaches
The research identifies regional differences in the areas receiving investment. Respondents in Europe and Asia placed greater emphasis on protection and compliance, while American businesses were more focused on process optimisation and product innovation.
These distinctions may reflect different operating environments, regulatory requirements, and stages of technology adoption. However, the published findings do not provide sufficient evidence to isolate the influence of those factors or establish why one region reported a particular priority.
There are also limits to the geographical comparisons. The survey included respondents from the UK, Germany, Spain, the Netherlands, Italy, France, the United States, Hong Kong, Singapore, and Japan, with around 100 participants in each market.
Individual country samples are therefore relatively small when compared with the number of large businesses operating in those economies. Their responses provide an indication of reported priorities, rather than a complete assessment of enterprise technology spending in each market.
For network providers and technology integrators, the findings point towards demand for services combining connectivity, security, and operational resilience. Supporting AI may require changes to existing networks, but the scale of investment will depend on the applications being deployed and the condition of the organisation’s current infrastructure.
Survey findings require careful interpretation
The research was conducted by Intuit Research among IT decision-makers and influencers responsible for digital infrastructure services. Participants worked for organisations with annual revenues of at least €500 million and more than 1,000 employees.
Consequently, the findings primarily describe the priorities of relatively large enterprises with established technology operations. Smaller businesses may have different infrastructure requirements, financial constraints, and relationships with external service providers.
The percentages also describe responses to survey questions rather than measurements of network utilisation, security effectiveness, or actual financial expenditure. A respondent identifying cybersecurity as a concern does not establish that their organisation has suffered a security incident or that its current controls are inadequate.
Similarly, respondents expecting technology budgets to increase may subsequently change their plans in response to financial conditions or other business priorities.
Colt’s results nevertheless show that security, resilience, and connectivity are prominent considerations among organisations introducing AI. The proportion prioritising additional capabilities has declined, while the operational demands associated with existing deployments continue to influence technology planning.
The company’s next research cycle may establish whether that shift persists, although the effectiveness of the investments being considered will ultimately require evidence beyond executive expectations and reported priorities.












