Telstra Technical Director Kim Krogh Andersen stated that before telecommunications operators pursue AI economy, they must first resolve fundamental issues such as network and operational capabilities.

Andersen is the executive officer of the Telstra group, responsible for networks, products, and technology. At the recently concluded FutureNet Asia event, he stated that telecommunications service providers are not yet ready for the AI economy, with the main gaps being in networks and operations.

He spoke frankly about the current situation of his own company, stating that the latest internal assessment by Telstra showed that the company was below level 2 on the TM Forum autonomous network framework. This framework is divided into 6 levels, from level 0 to level 5, and is used to measure the progress of telecommunications companies towards complete autonomy.

"It's a bit disheartening to start talking about intent-based slicing before getting the foundational work done," said Andersen. He believes that if the industry wants to reach level 4, it must abandon legacy systems and simplify and modernize networks and operations.

“AI will take advantage of vulnerabilities, which are those foundational tasks that we didn’t do well in the past,” he said, and pointed out that cutting-edge models can identify and exploit these gaps at machine speed.

Andersen believes that telecommunications companies should focus back on their core business, which is the network. “Our strategy is to return to the core and grow and innovate from there. No one on earth can convince me that the computing and storage in data center racks are more complex than connectivity. That’s not the case. But we have already lost confidence.”

He stated that the industry should adopt a product engineering mindset and break away from the habit of selling products in megabytes and gigabytes, rather than focusing on the results achieved.

"The future network will be fundamentally different. We cannot wait for 5 or 10 years to get ready; we must lay a solid foundation now. We need to view the network as an always-on connectivity capability that supports intelligent agents, devices, and autonomous workflows in society, because in a world driven by AI, connectivity will support far more than just communication," said Andersen.

"The internet is our product, and we need to treat it as such. This means adopting a product engineering mindset, continuously improving network capabilities, and managing them throughout their entire lifecycle, rather than only dealing with them after the devices are deployed and support for them ends. It also means addressing issues with legacy infrastructure, because we cannot build the level of automation and intelligence required to drive the future on a fragmented and outdated foundation," he added.

Comments on Andersen appear against the backdrop that while global telecommunications companies are facing continuous growth in connectivity demands, they still struggle to increase their revenues. Although service providers need to invest substantial capital to keep up with the increase in traffic, revenue growth remains sluggish in most markets. On the other hand, ultra-large cloud vendors and content providers are able to derive value from this traffic growth.

He compares service providers with cloud service providers, the latter of which earn more as usage increases. He says this is mainly because telecommunications companies are still “inherently localized” and unable to replicate global strategies. Therefore, he believes that API is the best way for telecommunications companies to monetize their network capabilities.

"We are building API. We have already deployed 6 API to ensure that application providers can start accessing these capabilities. We must begin to monetize these capabilities, and as AI matures further, these capabilities will also become more complex," explained Andersen.

He also stated that the AI traffic pattern is changing network demands. Upstream traffic is growing faster than downstream traffic, and latency between data centers as well as between data centers and end-users is very important. Andersen believes that telecommunications companies need to establish business models around these patterns.

Regarding Telstra, as Andersen mentioned, last year the consumption of token increased by 275%, and currently it reaches an average of 15 billion token per day, covering over 400 use cases, while costs have decreased by 7%. He pointed out that 56% of token is used for the customer experience layer, which includes Telstra's first AI voice agent for customer calls; software engineering accounts for 24%, autonomous networks account for 12%, and network security accounts for 8%.

He also mentioned the joint venture between Telstra and Accenture, stating that this allows Telstra to have "a governance perspective to manage and optimize its AI ecosystem responsibly. It shows which systems are running, how they are performing, and where risks or costs may arise, thus giving Telstra the opportunity to take early action and make data-driven decisions."

Andersen calls on the entire industry to “not expect someone outside to come and save this industry. We have to rely on ourselves.”