AI success depends on network visibility as enterprises struggle with growing complexity

Tyron Silk, Senior Solutions Architect, CASA Software.

---

AI can be said to be dominating business agendas and strategic discussions, but new research by Broadcom reveals that the biggest obstacle to successfully deploying AI is in fact the network supporting it.

This is according to CASA Software – which positions itself as a leading South African digital transformation organisation that partners with global technology leaders, including Broadcom, to enable its customers to realise the value of AI-driven operations and streamlined observability.

Broadcom’s 2026 State of Network Operations research reveals that while organisations are rapidly embracing cloud-first strategies and planning AI adoption at scale, most enterprise networks remain ill-equipped to deliver the visibility, performance and resilience required to support AI-driven operations.

According to the report, 99% of businesses have adopted cloud strategies, with hybrid cloud emerging as the dominant deployment model. Despite this widespread digital transformation uptake, fewer than half believe their networks are fully prepared for AI. Network congestion, latency, limited visibility, security implementations such as zero trust networking and SDWAN, increase complexity and continue to undermine performance. This creates significant risks for seeking to scale AI initiatives.

The research highlights a growing disconnect between AI ambition and network readiness. For example, 92% of organisations are reported to be planning to deploy AI-enabled networking solutions, but only 23% have implemented them, with 70% remaining in the early stages of network automation.

Tyron Silk, CASA Software Senior Solutions Architect and a Broadcom Software Knight in this field, notes the race to adopt AI has shifted attention towards applications and data, with many organisations overlooking the critical role the network plays in AI success. "Without complete visibility across increasingly complex hybrid environments of both the managed and unmanaged networks, AI workloads can quickly be affected by latency, congestion and unpredictable performance."

Visibility has emerged as one of the defining requirements for modern network operations. Nearly every enterprise surveyed in the report identified improved network observability as essential to supporting AI. “The highest priorities include real-time flow monitoring, user experience, infrastructure monitoring and enhanced network security. Visibility into external environments such as internet service providers (ISPs) and public cloud platforms has also become equally important,” says Silk.

He highlights that as businesses increasingly rely on cloud services, SaaS applications and distributed workforces, the network extends well beyond traditional data centres. “However, this expanded ecosystem also introduces significant blind spots. The Broadcom report found that 87% of companies experience limited visibility across internet and cloud environments, while 95% lack adequate visibility into key network segments, particularly within public cloud infrastructure,” adds Silk.

He notes these blind spots have a direct business impact. “Broadcom reveals that more than half of respondents cited visibility as essential for analysing network delivery issues. However, nearly half are reported to rely on it to predict capacity constraints, assess the impact of routing changes and monitor end-user experience. Improved visibility also enables faster root-cause analysis, proactive performance optimisation and reduced mean time to resolution (MTTR), allowing IT teams to resolve issues before they affect employees or customers,” he says.

According to Silk, despite these priorities, many network teams continue to face operational pressures that limit their ability to modernise. “Security remains the leading challenge, followed closely by dependence on other IT teams, managing multiple ISPs, constrained budgets, skills shortages and inadequate networking tools. These challenges have contributed to growing reliance on third-party providers, with 76% of organisations outsourcing elements of network operations.”

The Broadcom research suggests that AI itself may become part of the solution.

Silk says AI-enabled network operations platforms are increasingly being viewed as a way to automate routine operational tasks, identify performance anomalies faster and provide intelligent insights that improve network resilience. “Rather than replacing network engineers, these technologies are expected to augment existing teams by reducing repetitive work and enabling skilled professionals to focus on strategic initiatives.

“One solution helping businesses address these challenges is Network Observability by Broadcom, which provides end-to-end visibility across hybrid, multicloud and enterprise network environments. The platform delivers comprehensive monitoring across on-premises infrastructure, cloud platforms, SaaS environments, SD-WAN and SASE architectures, while also providing visibility into external routing decisions made by internet service providers and cloud providers.”

By combining AI-enabled analytics with deep network observability, companies can identify performance bottlenecks across every stage of the network delivery path, isolate root causes more quickly and proactively detect issues before they impact business operations. “This unified approach enables IT teams to improve network resilience while supporting increasingly demanding AI workloads.”

Silk emphasises that the network has become the foundation on which every digital initiative depends. "As AI adoption accelerates, organisations can no longer afford fragmented visibility or reactive operations. Success will depend on investing in intelligent observability that provides a complete view across every user, application, cloud platform and network path."

Silk says the report findings reinforce a clear message for enterprise leaders: AI readiness is no longer determined solely by compute power or data quality. “Network performance, visibility and operational maturity have become equally critical factors in delivering reliable AI services and exceptional digital experiences. As enterprises continue their AI transformation journeys, organisations that prioritise network observability, automation and end-to-end visibility will be best positioned to unlock the full value of AI while ensuring resilient, high-performing digital infrastructure,” Silk concludes.