Wayne Lotter
Head of International Networks
Early discussions and predictions about artificial intelligence and automation inspired technologies that completely reshaped society.. Andrew Ng, a globally recognised AI expert, summarised the technology’s impact on nearly every aspect of life in 2017, when he said “AI is the new electricity,”.
Many early proponents of AI focused not on its danger to human intelligence, but rather its potential as a tool to improve and support it. Former IBM CEO Ginni Rometty expressed clearly that AI would help us rather than replace us, pointing to a future where people and machines work together.
More than any other industry, telecommunications is pivotal to the AI revolution. As demand for connectivity accelerates beyond traditional boundaries, driven by cloud computing, AI workloads and the proliferation of edge applications, telcos face unprecedented challenges in network capacity, resilience and operational efficiency. For the international arm of Australia’s largest telecommunications company, Telstra, the solution lies in expanding infrastructure and fundamentally reimagining how networks are architected, operated and managed through AI and automation.
Wayne Lotter, Head of International Networks at Telstra International, has unique insights into this transformation. Leading Telstra’s international network engineering and operations whilst serving on the company’s global network and technology executive team, Wayne oversees the digital infrastructure and connectivity technology that spans subsea systems, terrestrial fibre networks and connectivity services across more than 30 countries.
As a result, he well understands how telecommunications providers must evolve to meet exponential growth in capacity demands whilst maintaining the security, resilience and performance that enterprise customers require.
The scale of the challenge is immense. Telstra International’s footprint covers
400,000 kilometres of subsea cables worldwide, while Telstra’s mobile network covers more of Australia than any other provider. Under Telstra’s new Connected Future 30 strategy, the company has positioned network investment and digital leadership as cornerstones of its future. AI integration serves as a critical enabler rather than a peripheral enhancement.
The architecture of autonomous networks
Telstra International’s approach to AI integration marks a clear departure from traditional network management. Rather than simply layering AI onto its existing infrastructure, it is implementing a network modernisation plan, placing AI and machine learning capabilities at the core.
As Wayne explains: “Our approach is to design a network architecture specifically built to support AI and machine learning capabilities. Rather than adding these technologies as an afterthought, we’re starting our roadmap by architecting the network around their requirements from the outset.”
The architecture-first strategy targets three essential outcomes that define network excellence: high resilience and reliability, robust security and the flexibility to meet evolving connectivity needs. Achieving these outcomes, however, requires more than smart algorithms. Rather, it calls for a fundamental simplification of network design.
Over the decades, telecom networks have grown increasingly complex, with multiple technology variants, vendor-specific systems and legacy integrations. Wayne compares the challenge to car manufacturing: “Think of an electric vehicle (EV) manufacturer—they typically build different models on a shared platform, using common components and electronics. The same logic applies, making standardisation key.”
Simplifying the network makes it more adaptable to AI-driven management, says Wayne, noting: “The simpler you make things, the faster and more reliably you can enable autonomy driven by data and rules.”
This approach means eliminating unnecessary technologies, standardising components and building shared platforms. Lastly, modernising the systems that control network operations allows AI to execute human-defined rules in real-time. The result is a hybrid model in which human insight shapes strategy while AI delivers agility and efficiency.
Intent-based networking: meeting hyper-scaler demands
AI-driven network management is especially valuable in intent-based networking, addressing bottlenecks in traditional service provisioning. Rapid growth demands faster deployment for hyper-scale clients, like cloud providers and financial firms. “They’re looking to expand their networks at a rate each quarter at 30, 40, 50% growth,” Wayne observes, explaining how manual provisioning processes can’t meet the need for speed and scale.
Intent-based networking addresses this challenge by enabling customers to specify desired outcomes rather than detailed technical configurations. “A customer or somebody who’s designing that solution can specify meeting outcomes and the network will be able to respond to those outcomes either by providing a design or configuring the network so that it can meet those particular outcomes, those particular characteristics,” Wayne explains.
This approach transforms the relationship between the service provider and the customer. Rather than engaging in lengthy technical discussions about routing protocols, bandwidth allocation and redundancy mechanisms, customers can define performance requirements, security parameters and connectivity objectives. The network then autonomously determines optimal configurations and routing to meet these intentions.
The resilience component of intent-based networking provides additional value for mission-critical applications. In subsea network operations, cable failures represent inevitable operational realities. Traditional recovery procedures require manual intervention, network rerouting and customer notification processes that can extend service disruptions. Intent-based systems enable automatic failover and self-healing capabilities that maintain service continuity without human intervention.
Multi-cloud integration and hybrid environments
The evolution of enterprise computing architectures toward multi-cloud and hybrid environments creates additional complexity for telcos. Applications and services are no longer concentrated in single data centres or geographic locations. Instead, they are distributed across multiple cloud providers, edge computing locations and hybrid infrastructure deployments.
The shift toward distributed infrastructure is also reshaping network requirements. “We are no longer designing networks for a single, specific data centre,” Wayne explains. “Rather, the network must adapt to a wide range of data centre needs – as application usage and AI adoption grow, so does the number of data centres we need to reach.”
As AI adoption accelerates, connectivity demands skyrocket. Modern applications use distributed architectures, with services like authentication, processing and content delivery spread across regions. Manually managing the process is not feasible. “You need a highly autonomous, AI-driven network to handle the scale and pace of demand,” says Wayne.
The growth rates that telecommunications providers encounter in serving these distributed architectures are unprecedented, he adds: “Most telecom providers will be seeing 30 to 40% growth in their capacity requirements to keep up with their customer demand from AI and cloud.”
This growth shows no signs of abating, with AI adoption driving a continual acceleration in network demand.
Operational challenges and tool rationalisation
Introducing AI and automation into network operations raises concerns about tool sprawl and the increase in complexity. As AI technologies evolve, a growing number of vendors offer specialised tools for network monitoring, predictive maintenance, security analysis and performance optimisation. While these tools promise benefits, adopting too many point solutions can lead to fragmented systems and add to operational overheads.
Telstra’s strategy focuses on simplifying and integrating its toolset rather than expanding it. “Our goal is to reduce the number of tools we use and create a common framework that allows them to work together,” explains Wayne.
Relying on multiple disconnected tools often requires further integration layers, which can undermine the efficiencies automation delivers. Instead, Telstra prioritises identifying a core set of tools that address essential network management needs, supported by a unified orchestration layer that ensures they operate cohesively.
“Rather than piecing together several tools that barely work together, we focus on using a few well-integrated tools with a shared orchestration approach,” Wayne explains.
The orchestration layer is key to enabling AI-driven network management. Rather than needing AI systems to interact with many different tools, they connect to a single orchestration platform that manages tool interactions behind the scenes. “The AI does not need to understand every tool; it just sends instructions to the orchestration layer, which handles the execution,” Mark adds.
Importantly, Telstra’s tool rationalisation strategy does not exclude the adoption of productivity-enhancing AI solutions. Wayne states: “We use tools like Microsoft Copilot to improve operational workflows and customer service.”
Such solutions provide immediate, practical benefits while aligning with Telstra’s broader goal of streamlined, AI-enabled network automation.
Edge computing and ultra-low latency applications
The rise of edge computing adds new demands for AI-driven network management. Modern applications now rely on multi-tier architectures spanning regional hubs, edge sites and far-edge deployments at the neighbourhood or enterprise level.
Telcos must understand the evolving topology, explains Wayne: “There are effectively three types of data centres. Regional hubs like Frankfurt, Tokyo or Sydney; edge locations such as Manchester or Birmingham; and far-edge sites within city neighbourhoods.”
To meet user expectations for consistent, low-latency performance, networks must manage traffic dynamically, something manual operations can’t handle at scale. AI optimises real-time traffic by calculating the most efficient routes based on current conditions. “We use computation methods to determine the best traffic paths,” Wayne says. “Over time, AI and machine learning help adapt and improve traffic flow.”
The capability is critical for ultra-low latency applications like financial trading, real-time gaming and industrial automation. Each action demands minimal delay and highly stable and predictable network performance, even under shifting conditions or infrastructure failures.
Industry evolution and future outlook
Over the next 12 to 18 months, AI adoption in telecommunications will accelerate rapidly, reshaping infrastructure and applications. Wayne highlights three major trends driving this transformation.
First is the rise of agentic AI, systems where multiple AI agents collaborate to solve complex problems. “We’re seeing the evolution of AI working with AI to achieve outcomes,” he notes, signalling a shift from isolated use cases to integrated, multi-agent ecosystems.
Second is the growing importance of high-quality data. Although data has always been valuable, AI-powered networks rely heavily on its accuracy, completeness and availability. “The importance of data is escalating. Everything will depend on how well enterprises can use their data,” Wayne notes.
Lastly, robust digital infrastructure like subsea cables, terrestrial fibre and data centres. It is essential to support distributed AI and cloud computing. “This plumbing enables these vast opportunities,” says Wayne.
Surging demand continues to outpace infrastructure growth, with power availability emerging as a key constraint. Integrating AI and automation in telecommunications marks more than a technical shift; it requires fundamentally rethinking how networks are designed, operated and managed.
The advancement of AI is unstoppable. In 1959, British mathematician and cryptologist I.J. Good predicted: “Within a generation, I am convinced, few compartments of intellect will remain outside the machine’s realm – the problem of creating ‘artificial intelligence’ will be substantially solved.”
While it has taken more than one generation to reach this point, Good’s vision quickly materialises. Today, AI and automation touch nearly every sector and region globally. With progress accelerating, it’s not hard to imagine that the next generation will see growth far beyond what even Good anticipated.
To read the full article in the magazine, click HERE.