Network infrastructure will determine the fate of AI

Rami Rahim will be delivering the keynote address at HPE Networking Day in Sydney on 7 October 2026.

While there has been much focus on the models shaping the AI era, one cannot overlook a key system underpinning these complex workloads: the network.

The network of today is no longer background infrastructure. It has become a fundamental enabler of AI readiness because, no matter how advanced the AI model, without a robust network supporting it - securely connecting people, devices, applications, clouds, data and AI agents - AI at scale will fail.

In Australia, enterprise adoption of AI is on the rise. According to the Australian Bureau of Statistics, large businesses increased their use of AI roughly four-fold between 2022 and 2025, placing growing demand on the infrastructure required to support these workloads. As AI projects extend further across business operations, leaders need to consider whether their infrastructure is capable of supporting them.

A key priority for enterprises investing in AI capabilities has understandably focused on maximising the performance and utilisation of costly GPU infrastructure. Yet, AI is creating challenges that extend beyond compute. IT leaders must also contend with new demands around moving data across distributed environments, maintaining visibility and control, and securing growing numbers of users, devices and AI systems. Indeed, the changing role of the network is elevating network architecture from a technical concern to a strategic one, influencing not only performance and security, but also the token economics of how AI workloads are routed to the most appropriate model.

Complexity has outgrown human operations: the rise of the self-driving network

As networks generate more telemetry, events, and operational signals, the next phase of networking requires infrastructure that is self-driving, capable of observing its environment, understanding context, anticipating issues, recommending actions, and autonomously resolving known problems.

Australian networks are facing rapidly expanding demands as IT estates become more complex, yet IT teams are not growing at the same pace. Self-driving network capabilities offer welcome relief, improving automation and network defense capabilities, reducing the manual strain of routine operational tasks, and freeing up IT teams to focus on innovation and higher value work. Without this level of automation, organisations may struggle to keep up with the volume of data movement, application interactions, and operational complexity that AI introduces across modern networks.

The network is becoming a security control point

AI has dramatically altered the security equation. The same technologies helping organisations identify vulnerabilities faster can also help attackers discover and exploit them more quickly, reducing the time available to detect and respond to threats. With Australian enterprises embedding AI systems more deeply across their business operations, security can no longer be treated as a separate layer wrapped around infrastructure.

In the face of this growing challenge, the network is uniquely positioned to provide visibility into the devices, connections, and behaviours that underpin every AI interaction. Rather than simply moving traffic, the network has become an active enforcement point, helping organisations identify anomalous activity, apply policy consistently, and contain threats before they spread.

Getting this right will be critical for enterprises scaling AI. As users, applications, data and AI agents become increasingly interconnected, organisations will need networks that combine continuous visibility, Zero Trust principles, and integrated security to support AI at enterprise scale.

The biggest misconception about AI readiness

One of the biggest misconceptions surrounding AI is that readiness is primarily a technology challenge, rather than an operational one. Access to models, accelerators, and cloud services is important, but only makes up part of the picture.

That is why AI readiness needs to be approached as a whole-of-stack challenge, connecting compute, storage, data, security, cloud, and networking into an operating environment that can support AI from experimentation through to enterprise-wide deployment.

Technology solutions aside, we cannot lose sight of the fact that, in order to scale AI securely and reliably depends on the network. Leaders must have a clear understanding of where data resides, how it moves, who can access it, how decisions are governed, and how performance is maintained as adoption expands.

Australia’s AI opportunity will be won in the infrastructure

The conversation around AI largely focuses on models, chips and applications, yet without adequate attention on the underpinning infrastructure required to operationalise them at scale, these transformative technologies will be constrained. The network is no longer simply infrastructure that supports AI. It has become the platform that enables AI to operate securely, reliably and at enterprise scale.

Australia's growing investment in AI infrastructure presents a significant opportunity, and will require a network capable of delivering performance, resilience, visibility and security that enterprise AI requires.

The organisations that succeed with AI may not necessarily be those with the most advanced models, they will be those that take a systems-level view of AI, connecting infrastructure, data, security and operations into a cohesive whole.