The adoption of AI in network environments is taking longer to gain traction than many organizations had anticipated. At the same time, demands on infrastructure are rapidly increasing.
This is evident from IDC’s “AI in Networking 2026” study, which reveals a clear gap between plans and execution. ComputerWoche reports on this.
Many companies have taken steps with AI in recent years, but remain stuck in limited applications. The transition to broader deployment is proving difficult. IDC Research Director Mark Leary notes that organizations have barely expanded their use. Even companies that seemed further along have not made significant progress.
AI is deployed within networks in roughly two ways. On the one hand, to make infrastructures suitable for AI workloads; on the other, to make network management more efficient. In both cases, scaling up proves more complex than anticipated.
Security is a key factor. According to Brandon Butler, Senior Research Manager at IDC, the importance of AI in security is growing, partly because attackers are using the same technology. Organizations are therefore increasingly using AI to detect threats. At the same time, security remains one of the biggest obstacles.
In addition, integration issues play a role. Existing systems do not always integrate well with AI solutions, which slows down implementations. There is also a shortage of specialized knowledge. The study shows that 81 percent of companies are therefore allocating more budget to managed service providers.
Capacity is reaching its limits
Meanwhile, pressure on network infrastructures is increasing. For example, 89 percent of data centers expect to increase their bandwidth by at least 11 percent in the coming year. Connections between data centers are also growing, with 91 percent anticipating similar increases.
Growth is even higher in cloud environments. Companies are counting on an average of 49 percent additional bandwidth for cloud connectivity. The cloud thus remains an essential part of AI strategies.
AI at the edge is also gaining ground. Currently, 27 percent of organizations run workloads at the edge, while 54 percent plan to do so within two years. Butler observes that organizations with a mature AI approach are particularly moving workloads to the edge, indicating a broader shift.
The impact is clear. Bandwidth requirements at the edge are expected to rise by 51 percent. At the same time, the complexity of network management is increasing as workloads spread across data centers, the cloud, and the edge.
Expectations regarding AI are also changing. For example, 46 percent prefer systems that perform network actions autonomously. Another 41 percent opt for decision support, while 13 percent do not want to use AI.
According to Butler, this is linked to increasing complexity and a shortage of specialists. As a result, automation is becoming increasingly necessary.
It is striking that companies have less confidence in platform-based approaches. Instead, they are more likely to opt for specialized solutions that better align with specific needs.
According to the study, the main benefits of AI lie not in cost savings, but in improved IT services and efficiency. For instance, 31 percent cite improved service levels as the main benefit, while 30 percent point to efficiency gains.