Omdia view
Summary
Insourcing by telcos is picking up pace, especially in connection with AI and data analytics-related skills. Omdia’s 2024 Service Provider Digital Transformation Strategies Survey Report suggests that the insourcing of technology skills is the supplier and partner relationship change that telcos are currently making most rapid progress with. Over 60% of telcos claim that they are currently insourcing to boost their internal IT and development capabilities, and a further 30% are considering taking this route. As Vodafone CTO Scott Petty recently pointed out, telcos are “becoming software companies and we’re all insourcing.”
Insourcing to address the AI skills gap
The current insourcing drive is a clear response to the telco IT skills shortage that the telco industry has been experiencing over the past few years, along with re-skilling, training, and talent acquisition programs. This is because limited in-house IT expertise, especially AI-related expertise, continues to be one of the key challenges to implementing a successful telco digital transformation. The ability to attract and retain AI talent with relevant qualifications and/or experience is a critical factor not just to deliver AI-enabled efficiencies, but also to support new commercial AI operations.
The same survey suggests that AI expertise is the biggest skills shortage, although telcos believe they also need to expand the number of cloud architects, and cybersecurity experts. AI skills shortages were considered a major concern for at least one third of the telcos responding to this survey.
This is reflected in what we see happening in the market. The latest edition of the Omdia AI Skills Tracker, a dataset that tracks job postings specific to the development of AI solutions, suggests that the specific AI roles that telcos may be most recruiting for this year are data engineers and data scientists.
Customer engagement, network management, and analytics are the areas where there is currently the most telco adoption of AI, and there are an increasing number of telco domains where AI-related tasks will require support. Just to take one example, in the RAN domain operators will need to implement training and recruitment to ensure that RAN engineers are able to take on AI-related tasks such as model management and fine-tuning. There are also initiatives by operators to position themselves as providers of AI services to their enterprise customers. In some cases, more ambitious telcos are looking to build LLMs, but even when they are just adapting existing pretrained models, there is a need for the skillsets to work with retrieval augmented generation (RAG), prompt engineering and fine tuning to get maximum value out of generic models.
However, it is worth remembering it is not just about acquiring the deep skillsets required to support specific network management or customer engagement AI use cases or internal telco operating efficiency gains; it is as much about extending more basic skills across the telco business as a whole.
For example, Orange is investing heavily in upskilling its workforce across the board. According to its Chief Technology and Innovation Officer, Bruno Zerbib: “We’ve been training, we have about 40,000 people that have been using our AI capabilities as tools within the company—an incredible amount of investment in terms of training.”
Similarly, Vodafone Group recently expanded an agreement with Microsoft to deploy its 365 Copilot AI software with up to 68,000 of its approx. 100,000 employees. According to Vodafone CTO Scott Petty, the company is using Copilot to improve “operational efficiency inside the organization; rewiring the business to provide an enhanced customer experience; and unlocking growth opportunities through new products and services.” For example, customer care agents are having AI built into their day-to-day interactions, including live recording and transcription and checks on the quality of customer interactions.
According to Scott Petty, the company’s recently announced GenAI investments are as much about making daily tasks and processes more efficient and enabling a pervasive culture of business and process optimization—which by its very nature requires company-wide skills development that is not just focused on technical roles.
An ever-changing telco supplier relationship
The current telco insourcing drive is a logical response to a long-standing problem faced by the telecom sector. However, it is not as straightforward as it may initially appear. Insourcing skills can be costly at the best of times, and it does not help that the AI talent pool is not necessarily expanding as fast as demand. So, telcos are not finding it easy to recruit, train, and retain highly skilled staff.
Competition is fierce, not just with other vertical industry sectors but also with major big tech players like AWS, Google, and Microsoft. For example, Microsoft made the enormity of this issue explicit in last year’s annual report, where it cited the challenge of attracting AI talent as a risk factor for the company. This despite the inherent advantages of big tech players like Microsoft, which are not only considered attractive employers but also have the financial resources to capture top AI talent by offering generous salaries, extensive benefits and in some cases, lucrative stock options. For example, a recent job posting by Microsoft for an Applied Scientist in its AI Development Acceleration Program lists a salary range of up to $160,200 per year for a position considered an entry-level graduate role. Big tech players have also been using targeted acquisitions and strategic deals to bring talented AI workers into their ranks, so it is not just their ability to offer attractive compensation packages that will continue to concentrate talent.
To a limited extent, telcos can try to benefit from playing the same game as big tech. For example, Vodafone CTO Scott Petty recently stated that one of the anticipated benefits of Vodafone being an early adopter of GenAI is that it will help attract right talent to the organization, “making Vodafone an employer of choice.” However, the insourcing of skills can be expensive, which is why often only the major telcos can afford to adopt this kind of approach.
But not even the major telco players will not be able to go it completely alone. The relationship with vendors and other industry partners will have to change as big telcos with techco aspirations bring more software development and AI skills in house, but even the most powerful telcos will need to partner to access the support of vendors, cloud providers, integrators, and other parts of the ecosystem.
Moreover, the insourcing versus outsourcing choice is not always as binary as may at first appear, especially if a telco is outsourcing to a supplier or partner that the telco is already deepening its relationship with. In such instances the lines between internal and external resources can become blurred. In the Service Provider Digital Transformation survey mentioned earlier, telcos reported that they are also making strong progress with adopting closer strategic and business-driven partnerships with key vendor partners—this was in fact considered to be the second-biggest supplier and partner relationship change after the insourcing of technology skills.
There are also other forms of partnering and collaboration. For example, as part of a big initiative to position Singapore as an AI hub, Singtel is increasingly collaborating with a wide range of technology, academic, and industry partners, as well as working with Nanyang Technological University and the National University of Singapore to co-develop AI skills training programs for its employees.
So, yes, it makes sense for telcos to insource and invest in training their staff to build up their AI expertise, but this needs to be done in parallel with a more comprehensive and strategic partnering approach. Judicious partnering is crucial partly because a lot remains to be done if telcos are to convince their enterprise customers of their AI ecosystem credibility; but just as importantly, telcos will need to draw on the expertise that will continue to reside elsewhere, and will have to compete for skills against big tech players with deep pockets.
Appendix
Further reading
Telecoms IT AI Software Market Forecast Report: 2024‒29 (November 2024)
AI for RAN: An overview and exploration of key use cases (October 2024)
Service Provider Digital Transformation Strategies Survey Report – 2024 (August 2024)
AI Skills Tracker 1Q24 Dashboard (May 2024)
Author
Kris Szaniawski, Research Director, Access, Software and Transformation
