The challenge of ensuring high-quality network service is a crucial concern for network teams, primarily driven by user complaints regarding service quality. A significant finding from a survey of 372 enterprises highlighted that out of 328 responses, “more capacity” was the top wish for improving networking, emphasizing its importance over advancements such as AI.
Most enterprises operate three main types of networks: the data center network, the WAN or VPN, and the worker LANs. Among these, the data center network garners the most attention, as around two-thirds of enterprises base their networking strategy on it. Any issues in this core area can jeopardize the entire network strategy.
In the realm of capacity, nearly half of those who acknowledged its importance reported having increased their data center capacity in the last two years, with 30% planning additional increases in 2026. Surprisingly, AI was only identified as a driver for expansion by 11% of enterprises. The leading reason for augmenting capacity was to alleviate complaints related to quality of experience (QoE).
A staggering 80% of enterprises noted that responding to QoE complaints is among the most difficult and costly tasks in network operations. These complaints are particularly challenging as they often lack immediate technical symptoms, requiring a deeper investigation that can be hampered by the transient nature of the issues. Reports indicated that congestion and latency are the root causes of over half of these QoE complaints, suggesting that increased capacity could have mitigated many of these problems.
Interestingly, while the data center network bears the brunt of focus, issues with VPN and worker-LAN performance appear more localized. Only a quarter of enterprises identified problems in those areas, indicating that challenges in the data center are more impactful on a broader scale.
From a cost perspective, many enterprises see SD-WAN as an effective and economical solution to boost capacity for remote sites, though they recognize the need for pilot testing to verify the reliability of broadband internet access in those locations.
As enterprises anticipate the impact of AI, a common belief has emerged that while self-hosted AI will indeed require additional bandwidth, this need will primarily relate to the data center. There appears to be a perceived unpredictability in AI’s data demands, which may reinforce the existing trend of companies investing heavily in data center capacity.
While concerns about AI-induced capacity needs are present, many enterprises report they would utilize AI’s potential more as a justification for expanding capacity rather than a current necessity. The use of AI in these discussions may have influenced companies’ investment choices, shifting focus to capacity-enhancing projects.
The trend of increasing network capacity is reflected in the broader equipment sector, where interest in white-box devices for capacity and flexibility is rising, with doubled consideration anticipated by 2026. This trend may also explain leading companies’ aggressive marketing of new technologies, such as Cisco’s push for its G300 chip, which is being positioned as a facilitator for AI, even though the core driver remains the need for enhanced capacity and improved QoE.
In summary, as enterprises strive to address QoE challenges, the push for increased network capacity—especially in data centers—remains the key priority, shaping not just operational strategies but also equipment investments in the networking space.
