Generative AI has had enormous influence on the telecom sector — as it has on many industries — since its groundbreaking launch two years ago. Today, the application of generative AI in telecom businesses has reached a vital point.
A report produced by Altman Solon and Kyndryl, and sponsored by Google Cloud Platform, revealed that 85% of telecom providers have at least one generative AI use case in production. Only 30% of all use cases are making it to large-scale production, however. The journey from initial concept to fully scaled implementation is proving to be a significant challenge for many businesses.
The report, “From Prototype to Production: Scaling AI in the Telco Industry,” stated that while global telecom providers are enthusiastic about the potential of generative AI solutions, they are struggling to make them work. The study said that only half of the use cases are meeting business goals.
In addition to the complexity of implementation, expenses are higher than anticipated, a fact noted by 80% of respondents in the study. In one example, a large telecom provider based in the U.S. experienced 30% to 40% higher costs than it had budgeted. The solution was a scaled generative AI implementation using a large frontier model.
The report detailed several cases that seem promising: A leading U.S. telecom company saw a 30% boost in network operations efficiency after implementing generative AI. A top European mobile network operator (MNO) cut agent workload by 60% to 80% using generative AI-powered chatbots.
Even with these achievements, numerous telecom companies reported a modest overall impact. Additional hurdles included underdeveloped data and governance practices, unclear business value outcomes, and insufficient top-level executive support.
Customer service is a chief area for experimentation with generative AI technology, and network domain is gaining interest. In customer service, telecom providers are using generative AI to develop customer insight, to provide call-center agents with information quickly, and to transform the customer experience in general.
In other areas, such as network, marketing, IT, and software development there also has been progress over the last year. The study, which included conversations with telecom executives, revealed a wide variety of use cases, “ranging from aiding in network operations across the field and at the NOC to supporting provisioning, configuration, installation, troubleshooting, fault management, and resolution workflows,” the report said.
Telecom providers are discovering various applications for generative AI in every part of their business. While the study showed that the main value was found in customer service, many top operators are broadly experimenting. One reported having 500 use cases in different stages of pilots and implementation, ranging from common ones like developer support and agent assistance to specialized telecom-specific large language models for network automation.
The report is based on surveys of more than 100 senior business leaders at tier 1 and tier 2 telecom services providers globally in the U.S., Europe, and Asia Pacific, as well as some in-depth interviews. The report includes a high level of detail and many specific examples.
Telecompetitor recently interviewed a Google Cloud executive, who painted an optimistic picture about the uses of agentic and generative AI for telecom providers.
