Australian telecom operator Telstra recently suffered a nationwide outage impacting thousands of its mobile customers. At its peak, approximately 45% of calls and data sessions were affected, with more than 600 emergency services calls encountering errors.

In a statement given to the Australian Senate, Telstra CEO Vicki Brady attributed the failover to maintenance on network time-protocol infrastructure. While Telstra’s maintenance team followed correct procedures, a GPS card failed upon network restart because an earlier design change had not been properly documented, with a required software update not installed to boot.

Telstra’s network faux pas is perhaps a good example of what EY analysts have called the material strategic and operational risk of technology debt. A recent report from the firm argued that modernization is key to solving telecom’s long-standing issues with accumulated customization, legacy dependencies, and poor visibility across complex operational environments. Specifically, the consultancy giant advocates an end-to-end strategy encompassing operations support systems (OSS) and business support systems (BSS), as well as cloud-native infrastructure, AI-powered automation, and digital service orchestration.

Paolo Canale, EY Americas technology leader for telecom, told this title that operators have begun to break down monolithic, aged systems into microservices – small, autonomous, and open-sourced components that handle specific business functions independently. While outages will always happen, the microservice mantra may help operators fully mine their networks for better monetization and network efficiency.

EY gave the example of an African operator suffering from tech debt, relying on "unsustainable" level hardware requirements and system adaptations to meet everyday business needs. Its solution was to employ a microservice architecture based on TM Forum's Open Digital Architecture (ODA), which espouses modular, cloud-native, and AI-ready building blocks to remove the specter of operation siloes. This was used to form a catalog-driven, digital OSS platform capable of intelligent provisioning, automating its order-to-payment cycle.

Underpinning the architecture was the greenfield deployment of a private cloud to host its OSS.

“Operators are bringing in the hyperscalers where it makes sense,” Canale explained. “They are looking for interoperability as the No. 1 requirement to work with them as they’re not necessarily ready to remove anything. … There is also a lot of trying out the AI tools out there because there is not yet a 100% focus on one tool. There are a lot of players there and it's so rapidly evolving so there is a lot of trying multiple systems out before committing.”

With hyperscaler and microservices on board, the operator in question reportedly reduced new contract signings from a week to 10 minutes, as well as seeing 40% less use of hardware resources, helping to reduce maintenance costs and energy consumption.

According to Canale, operators who want to get to this footing shouldn’t take a “big bang approach” because of the complexity of their monoliths.

“It's really an incremental approach where you define a roadmap, identify whether the function can be extracted from your monolithic basis or whether you need APIs to form an incremental middleware integration layer," Canale said. "It should be flexible enough to increment the interoperability of the different systems, and then build the roadmap on top of that, that could literally decompose this monolithic application into functions, and let them do what they are supposed to do.”

The approach can also benefit fiber-optic network providers, with the EY report citing one British altnet firm which reduced manual errors and developed an early network fault detection system through automation based on open APIs. The firm in question reportedly accelerated its testing processes by up to 100-times with a 300% increase in test capacity.

AI needs human intelligence

EY argued operators with outdated infrastructure may also be missing out on lucrative opportunities in AI analytics, automated service management, cybersecurity, IoT, SD-WAN, and edge computing.

Hillol G Roy, principal in technology consulting, expanded on the AI opportunity as a usage-based service type service opposed to a monthly recurring revenue (MRR) one.

“For this you’ll need different types of partners to onboard quickly for certain types of services.," Roy said. "You will probably construct or deconstruct services very quickly based on the need, and all those come down to foundational principles of security, flexibility, and interoperability in your network, then having the right modularity in your architecture so you can quickly slot those partners in or out."

IDC predicts the worldwide telecom and network API market will generate more than $6 billion in revenues per year by 2028. Vendors plugging this opportunity range from the obvious (Nokia, Ericsson, and the hyperscalers) to smaller players like radio access network (RAN) firm SynaXG.

The Singaporean vendor recently demoed a unified cloud-native platform AI-augmented RAN (AI-RAN) with Red Hat and Japan’s Orex Sai that leveraged TM Forum-compliant APIs to integrate OSS/BSS with edge AI applications. As Orex Sai’s CTO put it, this opened up networks to video analytics, drone control, and location-based services, helping operators potentially become more than “just connectivity providers.”

AI also underpinned Telefónica and Nokia pilots this year using common protocols and exposed network APIs. But the AI path arguably relies on more than just APIs and integration. EY’s research showed telecom ahead of the adoption curve, with roughly 85% of operators reporting some form of AI use. Yet fewer than 30% saw a clear return on investment (ROI).

EY traced this gap primarily to talent and mindset rather than to tools, arguing that for OSS/BSS transformation to succeed, operators should prioritize workforce alignment alongside technology upgrades.

“There is a lot of training going on, and one part of the training is upskilling existing people and getting people from other industries or other players within the ecosystem," Roy explained. "The second part, where we believe more momentum is needed, is also training on new ways of thinking: if you are using AI to do the same job but faster, you are probably under leveraging the true power of AI. So that's where we feel the next wave needs to be: with this power in your hand, how can you create new ways of working very differently?”

This view is shared by other analysts. In a separate conversation, Dallas Dolen, U.S. TMT industry leader for PwC, noted that opposed to being the laggards of popular perception, operators were first to jump onto AI-enabled chat for customer service, as well as automated ticketing to aid backend service on the network. This came naturally from the specialized knowledge inherent to operators thanks to the nuance of the capabilities and technology to run day-to-networking in telecom.

“You actually have a lot more broad technical talent within telco firms. … So it's not even a question of whether or not they can anticipate and deal with what's going to happen next, as it's more a question of strategy: how fast do you move to ensure that your engineers are in fact adapting and using this technology, for example,” Dolen said.

According to EY, the ultimate end goal with AI and microservices is an autonomous network as based on TM Forum's Zero-X framework, which espouses networks centered on zero wait, zero touch, and zero trouble.

"It's essentially the ideal state of network, where it works so flawlessly that you don't even know there is a network," Roy said. "In order to enable that you need a set of 'self-Xs', like self-activation, self-provisioning, and so on. A very quick example would be for an enterprise customer who has different profiles of users, there are different SD-WAN and network settings. Instead of them having to go through a large call center where someone is going through a manual on all the configuration, it can this be simple intent-based thing that you just type in a comment ... so then it goes through a series of decision trees behind and activates that customer.

"If it's that simple, the telco can outsource or give the enterprise administrator the ability to do all of that. So that's a new revenue-generating service, it's a great customer experience, and it makes the entire process a lot more efficient. That's the North Star vision."