Telstra’s Andersen: ‘We need to think about the network as the product’

Connectivity is very much back in fashion within some of the world’s biggest telcos. And it’s happening as network-softwarization and increasingly autonomous operations enable communications services providers (CSPs) to rethink how they make the most of their network assets.

One of the world leaders in rethinking connectivity is Telstra: It was working on network-as-a-service as far back as 2018. In tandem the company took a highly disciplined approach to building and depoloying a service-based IT architecture, called Telstra’s Reference Architecture Model (TRAM). Based on TM Forum’s Open Digital Architecture, TRAM now extends from IT through to data and networks.

Telstra’s aim is to have customers consume telco connectivity with the same ease and scalability as they do storage and cloud compute from hyperscalers.

“We need to think about the network as the product, and not build ... a static network with a price tag on top,” says Andersen, who is co-chair of TM Forum’s Autonomous Network Mission Board.

The first example of how this works in practice is Telstra’s Adapative Networks, which launched commercially in June 2025 for enterprise customers. It exposes network connectivity as well as “the telemetry and attributes on top of the network in a programmable fashion to our customers,” says Andersen.

“We don’t have a thousand different fixed [enterprise] products anymore,” he explains, adding that Telstra operates Adaptive Networks as a software product with regular releases of new features.

Learning from hyperscalers

Like many telcos Telstra has looked at how hyperscalers have benefited from operating platforms that are scalable, programmable and global. “If we nail these three things that’s really what matters, and that’s what we are doing with Adaptive Networks.”

Recent telco experience has shown that as network usage rises, “we don’t necessarily earn more money. We just invest more money,” says Andersen, adding: “We need to reinvent the business model to be scalable … so if people use it more, we earn more.”

When it comes to programmability, the hyperscalers “have used tech expertise to innovate from that core out, where in telcos, they have innovated around the core. So, we need to innovate [with] AI-first [and software-first] in the core of the network” Andersen says.

As for global reach, he advocates telcos behaving more like the banking industry and throwing their collective weight behind federated platforms such as Aduna. “We need to be autonomous by design, and we need to ensure we can both operate domestically and globally with an aggregator platform,” Andersen explains.

Remodelling B2B sales

However, Telstra’s focus with Adaptive Networks on selling scalable, programmable connectivity services represents a profound change for the whole business. Take, for example, enteprise sales teams, which historically have sold everything from routers to cloud services in response to requests for quotes (RFQs) or information (RFIs).

“Enterprise sales is a lot about creating responses to RFQs and RFIs from big enterprises. Adaptive Networks represent a fully agentic reinvention of that whole sales [process],” explains Andersen. “Now, we’re so focused. We are a product company, and our core product is the network.”

This means enteprise sales teams have to be “trusted advisors on connectivity to our customers. They need to demonstrate the value of network,” says Andersen. “It’s a big transformation for our sales team.”

Previously, enterprise sales teams may have spent 80% of their time on back-office work related to RFQs and RFIs. Now, “they could really be out there on the street with the customers talking about a product they’re proud of,” says Andersen.

The aim is to “maximize the utilization of the network and then not be obsessed if it’s in a direct or indirect relationship,” he explains. “Of course, we would like direct relationship where it matters and where it is the best option. But we need to also be open for other models.”

Indeed, one of Telstra’s “top eight priorities in agentic [AI] and in AI across the company is to reinvent our sales around its core product,” Andersen emphasizes.

Architectural discipline

Reinvention is made possible by architectural discipline across composable IT, AI & data and the network. “That’s where we benefit from the journey we have been on with Open Digital Architecture,” says Andersen. TRAM is designed to foster structured and smart technology decisions and to reduce complexity and duplication across the entire business. The architecture defines Telstra’s technology building blocks, and how they are composed and used. Having helped pioneer ODA, Telstra today is playing a leading role in informing TM Forum’s AI-Native Blueprint, and notably in the areas of Model-as-a-service and Data Products Lifecycle Management (DPLM).

In a future of agents and autonomous networks a strictly managed architecture will become even more important, Andersen believes.

He warns that “if you don’t get that composable architecture in place, if you can’t commoditize the LLMs [large language models], if you can’t [move] between clouds, if you can’t get rid of software that tries to rip you off, then you will not be able to put the run cost down.”

As a result, “the cost of AI will explode,” he says. “You cannot control the cost of all these players, they want the return on investment ... in AI, which is fair, but you don’t want every single thing you do and all the efficiency you provide to just be revenue in the “Magnificent Seven” books.”

He also empasizes that “you cannot have legacy. You cannot operate if you if you’re not cleaning up.”

Choose your partner

An important part of Telstra’s clean-up strategy has been to replace hundreds of transactional supplier relationships with a handful of trusted partners that deliver business outcomes.

For example, the operator has gone from 400 software system integration partners down to two: Cognitive and Infosys.

It has also reduced its number of data platforms from 80 to less than 20, with a plan this year to move to 16 platforms, before a further drop to just three.

This reduction helps Telstra retain a close grip on data ownership. “I think about data ownership now in the same way as I think about obligations: It’s very critical,” explains Andersen. “We don’t expose it. We don’t give the data to the application of software providers. We don’t give it down to the hyperscalers. They do the computing and storage. But we own the data.”

In addition, in 2025 Accenture and Telstra announced a seven-year joint venture (60% owned by Accenture and 40% by Telstra) to help deliver the telco’s data and AI strategy.

One of their developments to date is an AI refinery. This helps Telstra identify the best LLM for specific use cases and where to deploy agents as well as helping control run costs.

Laying the foundation to scale

As with other telcos, it is still early days for deployment of AI agents within Telstra, where they are so far helping in call centers, with software development and in reinventing B2B sales processes.

Andersen, who is co-chair of TM Forum’s Autonomous Network Mission Board, describes agents as “a massive, massive opportunity”, including within autonomous network operations.

Yet he also cautions that telcos need to be disciplined in laying the right foundations for AN networks, warning that the industry risks deploying AN in a way that is too use-case based and fragmented.

“We need to be very clear about what we need to get in place on each layer in our network to ensure that we can actually get to that autonomy at scale and not only in pockets,” Anderson says.

Indeed, Telstra wants to reach Level 4 AN “as fast as possible because when you get to Level 4, you get the basics in place. And then from there, you can actually start being specific on use cases,” advises Andersen.

“I would prioritize getting the fundamentals in place first, because we can see that now in our IT and software stack, in our data and AI, if you get these fundamentals in place, you can scale, not only faster, but also more sustainably.”