Can CSPs realize opex reductions through AI?

If the purpose of AI is to automate functions and processes that would otherwise be carried out manually, is it not reasonable to consider whether it will enable CSPs to reduce the number of manual tasks across their organizations and, by extension, the number of people needed to do them?

But amid CSPs' general exuberance over AI's potential, it is rare to hear a CSP business leader talk about it in terms of the potential for making sizeable cost savings.

There are good reasons for their reluctance to do so. First, it is not a positive message to for employees (although this, seemingly, was not enough to stop former BT CEO Philip Jansen from saying in May 2023 that technologies such as AI and automation would enable it to cut 10,000 jobs by 2030). But another reason is because CSPs do not have enough proof points yet to see how efficiency and productivity gains will actually translate into the loss of different roles.

Despite a reluctance to talk openly about cost savings and job cuts there is a general expectation among CSPs that AI represents the best opportunity to reduce opex levels.

Whereas capex reductions, or capex growth reduction, will result from slowing investment in 5G, or from delaying investment in 5G, AI will have a greater impact on opex – through people savings and with a specific focus on network operations where operators spent $254bn, equivalent to 14% of total revenues, in 2023. Other opex categories where AI offers the potential for savings on people include Sales and Marketing and General and Administrative, both of which have a high proportion of costs from employing people.

So, what might be a reasonable expectation of cost savings across these opex categories? In our newly-published research report Building an AI strategy: putting the foundations in place we explore different scenarios for potential opex cuts across the CSP business based on AI’s likely impact.

Operators do not report opex categories consistently but research firm MTN Consulting has built a set of data for the global telecoms-operator business based on the information that is available. MTN Consulting’s data shows that total global telco opex was $1.48tn in 2022, equivalent to 83% of total operator revenues. Capex was $329bn. In our report we identified those opex categories which have the largest proportion of costs coming from employees and where – based on our own research – there is the biggest potential to use AI. We have then developed different long-term cost-saving scenarios for each of these categories and applied them to the MTN data. In each case there is a bullish and a bearish opex-saving scenario.

While the scenarios produce some interesting results – a 9% total opex saving based on a bullish set of assumptions and a 1.9% reduction based on more bearish ones – the reality is that it is much too early to produce good forecasts. However, they do serve as a context or framework in which to consider the potential cost saving benefits of AI. They could also be seen as a challenge to the industry to ensure that benefits are realized.

Many parallels can be drawn between the excitement generated by AI today and the telecoms industry’s thinking when private and public cloud became a viable alternative to on-premise computing. Cost-saving was seen as an important driver for cloud migration and many of the early business cases made such assumptions. These days the enthusiasm for cloud transformation is greater than ever. But the case for cloud adoption is rarely made on cost grounds.

Whether AI can generate significant cost-savings for CSPs will be a function of its role in driving automation, particularly in network operations which is a major focus for CSPs. But automating a specific task or process or function does not necessarily lead to overall cost savings. The adoption of new technologies, new tools and new capabilities has been a constant in all of our working lives, freeing up time to do our jobs better, to be more productive and to undertake new tasks, Adopting new technologies has not made our roles redundant. Will AI be any different? In practice cost savings will arise when AI enables a CSP to eliminate entire teams or job functions, when staff retire and are not replaced , or when there is a mandate to cut costs with no specific technology or capability in mind to help spread resources more thinly.