Businesses rushing to implement artificial intelligence are encountering a major obstacle: their existing networks. Recent findings reveal that while almost all companies have moved to cloud-based approaches, fewer than half believe their networks are fully prepared for AI.
Networks fall short of AI goals
A Broadcom study found that 99% of organizations now follow cloud-first strategies, with hybrid cloud emerging as the leading model. Only 23% have deployed network solutions powered by AI, even though 92% intend to do so. This disconnect leaves most companies in the initial phases of network automation.
Tyron Silk, a senior solutions architect at CASA Software, explained that attention has shifted toward AI applications and data, leaving networks overlooked. “Without full visibility across complex hybrid setups, AI workloads suffer from delays, congestion, and unreliable performance,” he stated.
The issue extends beyond technology. Nearly half of the surveyed companies depend on network visibility to anticipate capacity limits and track user experience. Yet 87% face blind spots in internet and cloud environments, and 95% cannot monitor critical parts of public cloud infrastructure.
Visibility rises as a key concern
The study points to an unexpected situation: while 70% of organizations remain in the early stages of network automation, they are already feeling pressure. Security tops the list of challenges, followed by reliance on other IT teams, handling multiple ISPs, constrained budgets, and skills shortages.
Silk observed that the move to cloud services, SaaS applications, and remote workforces has expanded networks beyond traditional data centers. “This broader ecosystem creates significant gaps in monitoring,” he said. Those gaps directly harm business operations by slowing problem diagnosis and increasing resolution time.
Related: Electric bike reaches 65 mph off road
Broadcom’s Network Observability platform demonstrates one solution, delivering full visibility across hybrid and multicloud setups. It tracks on-premises systems, cloud platforms, SaaS environments, SD-WAN, and SASE architectures while also monitoring external routing choices made by ISPs and cloud providers.
This level of oversight goes beyond fixing problems. Companies using observability tools resolve issues faster and achieve better performance for AI workloads. The ability to manage networks effectively may determine whether AI adoption succeeds or fails.
Network teams once focused mainly on keeping systems running and ensuring enough bandwidth. Now they must also handle AI workloads within security models like zero trust and SD-WAN. The change has happened quickly, and many organizations are still adjusting.
Silk stressed that networks now form the backbone of every digital project. “As AI spreads, companies cannot afford fragmented visibility or slow responses,” he said. “Success will require smart monitoring that offers a full picture of every user, application, cloud platform, and network path.”
The study’s message is straightforward: AI readiness depends on more than computing power or data quality. Network performance, visibility, and operational readiness have become essential for delivering dependable AI services. Companies that address these areas will be best positioned to benefit from AI.
