Login
Register
Telefónica has already reached an average network autonomy level of 3.05 and is operating nine Level 4 high-value uses cases in Spain, Brazil and Germany. Nilmar Seccomandi David, Telefónica’s Director of Autonomous Network and Infrastructure, details the company's AN strategy and how it lays the ground for agentic AI.
Joanne Taaffe, TM Forum
How Telefónica plans to achieve Level 4 AN by 2030
During Telefónica’s Capital Market Days in November, CEO Marc Murtra, gave prominence to the company’s goal of achieving an average level of 3.75 in network autonomy in Spain, Brazil and Germany by 2028.
It’s a tall order, but one that is achievable, according to Nilmar Seccomandi David, who will play a determining role in achieving this ambition as Telefónica’s Director of Autonomous Network and Infrastructure.
Seccomandi’s confidence is fueled both by the company’s strategic approach to developing autonomous networks (AN) and the foundations Telefónica has laid in the four years since it began its AN journey.
To date, Telefónica has reached an average autonomy level of 3.05 in its latest mid-year assessment. In addition, it already has nine Level 4 high-value use cases across its operating companies in Spain, Brazil and Germany. It plans to reach an average of Level 4 by 2030. Level 4 AN represents a major shift from traditional human-defined automation processes to true autonomous decision-making. At Level 4, the network can self-manage, self-optimize and handle complex tasks.
“Our success is based on a holistic approach to structuring the program,” explains Seccomandi. “We include all processes we have, from planning to operations, and all domains we have: fixed network, mobile network, CORE, Telco Cloud, IP and transport"
Now Telefónica’s advances in AN have paved the way for agentic AI, which Seccomandi believes will be an essential spur for further automation.
“If you are close to AN Level 4, you are prepared to implement agents,” says Seccomandi. “AI agents [are] fundamental to achieving zero-touch operations. They are the translator of intents and coordinate automation across different domains ... and will bridge the gap between the high-level business goal ... and low-level actions.”
However, he warns that “if the level of autonomy is between 1 and 2, I recommend other techniques... if you’re just starting out, first understand what you want to automate, then select the most suitable technique to address the problem.”
Indeed, laying the groundwork for AN demands an architectural and orchestrational rigour that also serves agents.
Telcos “need to think about the final architecture to scale agents, with all the necessary layers – such as infra, AI models, data, development, observability, security, governance,” recommends Seccomandi. And although AI agent “pilots are important,” they should be supported be “a reusable framework that can be adopted across different domains.”
Telefónica is still in the process of deploying its architecture. However, it has made enough progress “to start implementing many agents,” says Seccomandi. These include AI agents to automate the operation, fault management and configuration of software-defined IP networks.
To ensure a holistic approach to AN deployment, Telefónica has established four workstreams, which operate in lockstep.
The first focuses on network innovation; namely the building of an open, disaggregated cloud-based network architecture. At the same time, Telefónica is phasing out legacy networks, including the group’s copper network, which has already been switched off in Spain.
Seccomandi calls the second workstream “the brain.” Here, teams are creating a platform for automated intelligent decision-making by building a federated data architecture and an orchestration model, and modernizing operational support systems (OSS).
Seccomandi describes the third workstream as “the heart,” because it involves rethinking and redesigning the company’s end-to-end processes lifecycle.
The fourth workstream focuses on people and culture and “is our most important,” says Seccomandi. But it is also the most complex area in which to effect change.
“Our biggest challenge has been shifting the organizational mindset,” he explains. “This requires redefining roles, introducing new technology, and fostering collaboration across traditional silos.”
An important aspect of Telefónica’s strategy is integrating AI into its operations and processes. To this end the AN workstreams are supported by the company’s AI Centers of Excellence (COE), which were established two years ago.
The Global AI COE serves as the mothership for local AI COEs in Spain, Brazil, and Germany), which implement the global guidelines. Its responsibilities include:
But although “the leadership and orchestration of the program is centralized, the responsibility of the program is decentralized,” says Seccomandi, and “network areas are involved in the decision to automate their processes.”
If, for example, an operating company is implementing a digital twin, it can turn to the COE for guidance on implementing CD/CT (continuous deployment/continuous testing) or creating a closed loop, as well as which databases to use for radio, fault management, performance or inventory information.
A fundamental aspect of Telefónica’s thinking is that the end game is creating “the best future network,” says Seccomandi, which is one that optimizes data usage, improves customer experience, supports hyper personalization, and enables innovation in services and service level agreements (SLAs) – all while maximizing capex and opex efficiency.
But specific goals are not dictating the paths to autonomous networking. Nor is the company seeking to hit 3.75 across every use case; instead the goal is an average.
“We are looking for new opportunities based on the network, but first we are thinking how we can increase network automation and then in parallel, how we can use this automation to provide new services for our customers,” says Seccomandi.
That doesn’t mean there aren’t immediate payoffs. Automation, for example, can speed up 5G site rollout. And "in Brazil, we are accelerating the deployment of 5G with autonomous use cases,” notes Seccomandi.
Saving money is another benefit. “When we created a new IP network, we were reusing 47% of the fiber. It meant we had to build 53% of the fiber. With autonomous use cases and intelligence, we are reusing 20% more fiber,” Seccomandi explains.
In the meantime, Telefónica will continue to “collaborate across the ecosystem on standards, partnerships and shared learnings, which are key to accelerate the programs,” says Seccomandi.
