BLOG | Building AI-native networks: A conversation with Kim Krogh Andersen

The AI era is raising the stakes for telecom networks. Telstra's Kim Krogh Andersen and TM Forum CEO Nik Willetts discuss what it actually takes to get ready — and why the industry can't afford to wait.

The rise of AI isn't just changing what telecom networks carry. It's changing what those networks need to be.

At DTW Ignite 2026 in Copenhagen, Telstra Group Executive for Networks, Product and Technology, Kim Krogh Andersen, sat down with TM Forum CEO Nik Willetts to cut through the noise on building AI-native networks and discuss why operators who hesitate risk being left behind in the fastest-moving super-cycle the industry has ever seen.

The network as product

Telstra's approach starts with a shift in mindset. In its connected future strategy, Telstra has been clear: it wants to treat the network as a product, not something static. It's a perspective borrowed from the hyperscaler playbook. The same way Amazon or Google think about a data center as the product — innovating from the inside out — Telstra thinks about its network.

Hyperscalers think about the data center as their product and prioritize innovation from inside out, not by building on top of it. Similarly, Telstra is focused on connectivity as its core and the network as its product. When you think about the network that way, you really want to innovate in the network. That's what autonomous networks is all about; taking all the AI capabilities out there and embedding them directly into the network. But, to provide network as a product, AI needs to be built in by design.

This isn't abstract philosophy. Telstra has structured its organization around it by bringing network, product, and technology under one roof, with a shared P&L. When the network is the product, AI by design isn't a feature add-on. It's the foundation.

A new kind of demand

The pressure to move is real, and it's building fast. More than $700 billion is flowing into data centers globally this year, with more to come. Those data centers will not be islands. They need to be connected, both to link data centers together and to bring AI from devices, from humans, from machines, to those data centers.

That changes everything about how networks must operate. AI agents run 24/7. They don't peak at 8pm and go quiet at 3am. Today, operators can take the network down for a change in the middle of the night because people are not using it. That won't work in a world where agents are up and running around the clock. When agents have this kind of always-on demand, operating procedures need to change and networks need to operate fundamentally differently. And, that cannot be delivered with a human-centric operating model. 

This shift also forces a rethink of the revenue model. The challenge of today's business model is that the downloaded megabyte is becoming an obsolete currency. If operators stimulate customers to use more of their product, they don't necessarily earn more money but they definitely invest more. In an AI future, that currency needs to change, ensuring operators are paid for the value the network delivers. In this case, reinventing the business model is critical.

Stop admiring the problem

Kim is direct about where the industry stands. It has a habit of analyzing its problems at length without actually solving them. His call to peers: stop deliberating and start moving.

There is no perfect playbook, and there never will be. But the industry has a tendency to admire the problem instead of doing something about it.

As chair of TM Forum's Autonomous Networks Mission Board, Kim is working with operators including China Mobile and Telefonica to build a practical benchmark and playbook. This playbook will give every operator a starting point, domain by domain, level by level. The goal isn't perfection, it's momentum. Get the data foundation in place, ensure the semantic layer is in place, and approach the most valuable processes with an AI-first mindset — not by layering AI on top of existing workflows, which won't deliver the results needed.

What progress looks like

Looking ahead, Kim's ambition is clear. He wants to see operators moving at pace and at scale to demonstrate end-to-end progress, with the foundations in place and results showing up on the P&L. The goal is autonomous networks moving away from individual use cases toward operators that have moved with pace and scale across the network, with the benefits becoming visible.

His vision for the industry is built on structured collaboration. Finding allies helps the industry move faster and with more precision than when we work alone. That's the work TM Forum's Autonomous Networks Mission exists to enable.

The playbooks, benchmarks, and solution packs are already being built. The question is whether operators will use them with the urgency the moment demands. Kim puts it plainly:

"In this AI race, you're by default behind and need to catch up. We need to move with conviction if we want to take the benefits of this AI super-cycle."

The future is AI-native. The race is already underway.

Watch the full conversation: Building AI-native networks: A conversation with Kim Krogh Andersen