AI in Telecoms: From Hype and Opex Savings to Far-reaching Business Impact

Cisco’s AI Readiness Index shows that organizations across sectors are racing to turn their AI strategies into actual deployments. Yet, many lack capabilities in the crucial foundations, such as infrastructure and data, to realize AI’s full potential. Communication service providers (CSP) have been among the early adopters of AI technologies, and there is a lot of excitement around the Opex savings that they could generate. But with the right building blocks in place, the business benefits of AI could extend well beyond operational efficiencies, driving top-line growth for CSPs too.   

AI has got to be the most hyped technology we have seen since the advent of cloud computing more than 20 years ago. This hype has led to an urgency in organizations across industries to adopt AI, as Cisco’s AI Readiness Index shows. Yet, the study also shows a big gap between many organizations’ AI ambitions and their ability to make the most of these technologies. Only 14% of businesses globally are fully leveraging AI-powered technologies, because some of the critical foundations, such as infrastructure and data, are not yet ready for deployment.

More than just driving efficiencies

A couple of months ago, I took part in a panel discussion on AI and automation at India Mobile Congress, looking at the challenges and opportunities around these technologies for communication service providers (CSP). Opex savings was highlighted as a key driver for adoption, but CSPs need and want more than just to cut costs. They want to future-proof their business and become more resilient. They want to boost customer satisfaction to reduce attrition and reduce time-to-market for new services. They hope that AI and automation will help them get there. So, while efficiency gains are often a byproduct of these technologies, I don’t believe they are the main driver for adoption amongst CSPs.

One thing that all panel discussion participants agreed on – and that the AI Readiness Index highlights too – is that CSPs’ underlying infrastructure determines their success in adopting new technology innovations such as AI. However, 54% of the Index respondents say that their infrastructure has only moderate or limited scalability and requires enhancements to handle complex AI applications. Also, only 26% have fully integrated tools with direct and automated interactions.

The power of self-X networks

AI and automation go hand in hand. AI can be an invaluable tool for turning network data into actionable insights – but you can’t automate what you can’t see. Insights are only as good as the available data allows. That’s why CSPs need three things: network telemetry data to make informed decisions; control points, i.e., automation platforms, to be able to act on those decisions; and a robust and high-performance network infrastructure to execute those decisions.

As the AI hype has gained momentum over the past six months, we have learned through conversations with our customers that they want to use these technologies to create self-organizing, self-maintaining and self-service networks that deliver benefits beyond Opex savings. After all, CSPs don’t make money running networks – they make money selling services.

Competitive differentiation

 Self-organizing and self-service networks also help CSPs become more agile and monetize their huge network infrastructure investments more effectively. Several weeks’ lead times for new services are reduced to hours or even minutes, and delivering services on-demand becomes possible too.

 The combination of AI and automation also makes networks autonomous and more robust, which allows CSPs to differentiate with ultra-resilient, low-latency services. One of our customers saw an increase in revenue of $25M a year by bringing new services to market faster and creating service differentiation with guaranteed SLAs. To pave the way for this transformation, CSPs need to ensure that their underlying transport domain network devices have the capabilities needed – such as Segment Routing that automation systems can leverage – to deliver those differentiated experiences to customers.

Meaningful cost savings

We have seen a few CSPs announce new innovative generative AI initiatives recently. For example, Lumen is piloting a Microsoft 365 Copilot tool to help customer service agents find the information they need and resolve issues faster. Similarly, AT&T’s generative AI tool Ask AT&T is designed to automate internal processes, boosting the productivity of coders and software developers, for example. The Opex benefit of initiatives like this is that they free up operations talent, allowing colleagues to move away from routine or repetitive tasks that are relatively simple to automate, and instead focus on finding solutions to bigger issues.

Self-maintaining networks with AI and automation helps CSPs generate Capex savings too by optimizing the use of network resources while ensuring a seamless end-user experience. One of our customers reduced Capex by $14M in this way. To achieve these savings, CSPs need to have the necessary control points and telemetry within their network, so that they have visibility into the behavior and performance of different services and predictive insights into where issues might arise. Software-defined network controllers in the transport domain are critical. They enable the creation of a network-as-a-platform that AI can use to understand the live characteristics of a network and how it reacts – and use this information to anticipate and avoid network issues. Furthermore, network assurance solutions such as Accedian Skylight – with automated data correlation, anomaly detection and predictive analytics – enable CSPs to pin-point performance issues in real-time and pre-empt how they might jeopardize quality of experience. This allows CSPs to move from reactive to predictive zero-touch operations, and cost-effective delivery of high-quality services at scale.

In the constant battle to reduce customer attrition, actionable network performance insights can be a hugely powerful tool for CSPs. We enabled one of our customers to use the control points and telemetry of a multi-layer optical and IP network and AI algorithms of previous failure characteristics to identify an optical link that was beginning to degrade. Thanks to understanding the relationship between the layers and the traffic profile of services riding over this link, the system rerouted the service to a different link before the link was set to fail.

As hype converts into adoption, CSPs need to ensure that they lay the critical building blocks to maximize AI’s full potential. And it is a race against time, as 61% of Cisco’s AI Readiness Index respondents believe that they have just one year to deploy their AI strategy before they will see a negative impact on their business. Opex savings might help CSPs make the case for AI technology investments, but today’s investments will drive future top-line growth too. From improving business resilience and reducing time-to-market for new services to enabling competitive differentiation, the positive impact of AI will be far-reaching.