The Questions Telco Execs and Experts are Asking about AI in the Data Centre and Transport Network

As AI traffic reshapes bandwidth, power and latency demands, telcos are racing to keep pace with hyperscaler-scale growth and use AI itself to help them do it.
Network X 2026 runs 13–15 October in Vienna. Register now. Operators and vendors will be there to talk about what's next for AI in the data centre and transport network.
Inside data centres and across the transport links that connect them, operators are now re-architecting power and automation policy in near real time, just to keep pace with hyperscaler demand.
The key questions that Communications Service Provider (CSP) executives are asking include: Can transport networks absorb AI-scale traffic growth without a costly rebuild? How much of that network is a CSP willing to let AI run and power down on its own? And can telcos capture a share of the AI infrastructure boom that today mostly flows to hyperscalers and their suppliers?
At Network X 2025 in Paris, Omdia's Ian Redpath, Research Director for Transport Networks and Components, opened the IP Transport & Network Convergence track with a picture of an industry mid-buildout. The top four cloud providers' combined capex was projected to near $350 billion for 2025, with Amazon alone at roughly $100 billion, and EMEA data centre capacity was on track to grow from 29 GW today to 47 GW within three years.
Fibre construction was picking up in ways the industry hadn't seen in a generation, and more than 500 new “AI shelters” are replacing amplifier huts that were never built to carry AI-era loads. Telcos such as Lumen are already installing them, Redpath said. AI is driving a vast ecosystem, he added, that runs from enterprise data lakes through the “AI middle mile” to gigawatt-scale training campuses, and a holistic view of it is now critical for anyone planning data centre and transport capacity.
Here's what the latest research and these telco conversations reveal about where AI in the data centre and transport network is actually paying off, and where operators would like to go next.
Execs and Experts Featured in This Article
Ian Redpath, Research Director, Transport Networks and Components, Omdia
Hayrullah Kolukisaoglu, IP Network Principal, Turkcell
Sterling Perrin, Senior Principal Analyst, Heavy Reading—read the survey findings
What does the research show about AI's impact on the data centre and transport network?
A Heavy Reading survey of 77 network operator executives, published in August 2025, put numbers on the gap between expectation and readiness (TelecomTV). Twenty-nine percent of operators expect AI to make up more than half of their long-haul traffic within three years, and 18% expect the same for metro traffic. Yet only 16% called their networks “very ready” for that demand, and a further 39% said they were ready but still had work to do. North American operators were notably less confident than the rest of the world (44% ready or very ready, against 68% elsewhere), even though they face some of the earliest AI traffic growth. Capex constraints, thin management buy-in and security concerns were the top-cited obstacles.
What Telco Execs Are Asking About AI in the Data Centre and Transport Network
Can the network handle the traffic?
Operators expect AI to dominate long-haul and metro traffic within three years, but Heavy Reading's research shows readiness lagging expectation, especially in North America.
Can operators trust AI to act on live infrastructure?
Agentic AI scores highest for troubleshooting and live optimization, but accuracy and reliability concerns mean most CSPs are staging autonomy with guardrails rather than removing humans from the loop.
Omdia followed up with its own survey of 80 qualified CSP respondents, fielded in November 2025 with Ciena and Cisco. It found AI already routine in daily transport operations at 12% of operators; another 24% expect to reach that point within a year, and 45% more within one to three years. Asked why, operators pointed to fewer human errors and faster troubleshooting (48% each) and lower opex (40%). Digital twins drew the sharpest interest as a specific technology, rated most valuable for network optimization and traffic engineering (46%) and for performance monitoring (43%) (Light Reading).
Where are operators finding real value in the network today?
Much of it centres on turning predictive traffic models into power savings. AI doesn't treat every router as always-on. Instead, it can target specific components, such as ports and fabric units, for localized energy savings. It analyzes traffic patterns to anticipate demand shifts, then moves hardware in and out of sleep modes before it's needed.
The architectural catch, as Turkcell's IP Network Principal Hayrullah Kolukisaoglu frames it ahead of Network X 2026, is guaranteeing that power reduction never touches latency, throughput or reliability. The SLA has to survive contact with the energy-saving algorithm.
Will operators trust AI to run the transport network without a human in the loop?
Sterling Perrin of Heavy Reading followed up with an Omdia survey on agentic AI specifically. It found 47% of CSPs expect agentic AI to become at least very important in transport networks, with the clearest use cases in troubleshooting assistance (55%), live performance optimization (50%) and on-demand natural-language reporting (45%) (Light Reading).
But 45% also named accuracy and reliability as their primary concern. The report put it bluntly: “The move from traditional AI to agentic systems requires a significant operational shift, and few network operators wish to remove humans from the loop entirely.” Most CSPs are staging autonomy with guardrails and clear intervention points. Few plan to hand over the keys outright.
What's next for AI in the data centre and transport network at Network X 2026?
Ian Redpath returns to Vienna to chair the track again. He opens 13 October with an Analyst Welcome & Market Outlook, then joins panels on The Future of IP-Optical Convergence and Disaggregation and Open Networking: Balancing Innovation, Cost Control and Risk. He closes out on 15 October with a session that tackles the question hanging over the whole ecosystem: Telcos and Hyperscalers: Who Captures the Value in the AI Data Centre Ecosystem?
Turkcell's Hayrullah Kolukisaoglu delivers an Operator Address, AI Predictive Dynamic Energy Saving, in the IP and Optical Transport track on Thursday, 15 October, 14:00-14:20. He'll dig into granular power management and predictive traffic modeling, and the SLA safeguards operators will demand before they trust it in production. The same afternoon, the Data Centre Networks track runs its own Operator Address, Achieving Sustainability in Data Centre Networks, and closes with a Panel Discussion, The Future of Data Centre Connectivity: 2025-2030 Vision.
Network X 2026 runs 13–15 October at VIECON, Vienna. More than 5,500 telecom leaders are expected, spanning mobile, fibre, Wi-Fi and data centre networking. View the agenda and register to hear how operators are answering these questions in practice.
FAQs: AI in the Data Centre and Transport Network
Where is AI improving data centre and transport network performance today?
Digital twins for traffic engineering and performance monitoring, and predictive power management that targets individual router components, are the two use cases operators rate most valuable right now, per Omdia's November 2025 CSP survey.
What's the biggest obstacle to broader AI adoption in the transport network?
Readiness. A Heavy Reading survey found only 16% of operators call their networks “very ready” for AI-scale traffic growth, and trust is the other gate: most CSPs want a human in the loop before agentic AI acts autonomously on live infrastructure.
Where can telecom leaders discuss AI in the data centre and transport network in person?
Network X 2026 in Vienna hosts an Operator Address on AI Predictive Dynamic Energy Saving on Thursday, 15 October, 14:00-14:20, in the IP and Optical Transport track. View the session and register to join the conversation.
About the Author
Christopher Lycett is the Event Director of Network X and has been since the event's launch in 2022.