The telecommunications industry is making a major shift in priorities. Over the past few years, 5G has been the priority as it was viewed as a transformative technology that had the potential to make telcos significantly more strategic to their enterprise customers. As important as 5G is, it has given way to artificial intelligence (AI), particularly generative AI.

A recent report published by Nvidia, State of AI in Telecommunications, underscores this growing trend where telecom companies are embracing generative AI to enhance customer experiences, streamline processes, boost productivity, and optimize network operations.

The report is based on a survey of more than 400 global telecom professionals directly involved in implementing and delivering AI technologies. The survey targeted a diverse group of participants, including companies that offer mobile, fixed, and cable services and their hardware and software suppliers. The survey collected feedback from professionals at different levels of seniority to capture a comprehensive view of the industry’s AI adoption. It was conducted between October and December 2023.

AI becomes integral to telecos

The report uncovered strong interest in AI, with 90% of respondents assessing, piloting, or implementing AI solutions. This high level of engagement indicates that AI is becoming an integral part of the telecom value chain. Many industry professionals believe AI will play a crucial role in their company’s success, with 56% of all respondents and 61% of decision-makers expressing this view – a 14% increase from the previous year.

This stark shift in opinion is good to see. A couple of years ago, in my conversations with telcos, they showed a bit of interest in AI, but had some healthy skepticism about its value. Today, cloud native technologies, software, security, and other trends have made operating a network significantly more complicated. So complicated that even the most seasoned network professionals can’t keep up – but machines can.

Nvidia also observed a significant shift towards generative AI and large language models (LLMs), marking them as the standout AI technologies of the year, with 43% of telcos currently investing in generative AI. Of those respondents, 57% use generative AI to enhance customer service, and 57% to make employees more productive. Additionally, 48% of telcos use generative AI for efficient network management, 40% for network planning and design assistance, and 32% for marketing content generation.

The adoption of LLMs will be interesting to watch as new ones are constantly being created. Success for a carrier requires choosing the right ones today, but also having the capability to quickly adopt new ones when required. If ever there was a driver to embrace openness and cloud-native architectures, this is it, as it helps avoid lock-in.

AI costs less, driving revenue growth

According to the data, 63% of all respondents implementing AI have seen costs decrease in specific business areas, with 14% noting reductions of over 10%. Meanwhile, 67% reported that implementing AI has helped increase revenue, with 19% of respondents indicating revenue growth of more than 10% in specific areas.

The survey didn’t delve into how the telcos are generating more revenue with AI. My assumption is that it’s creating the capability to cross-sell services more effectively. Typically, this isn’t something telcos do, but AI should make that easier. Long-term, it will be interesting to see if customers would look to their telcos for AI advice, creating an entirely new service and product stream. Right now, that falls to the hyperscalers, most notably AWS and Microsoft, as well as system integrators such as Accenture and World Wide Technology, but the door is open for the telcos.

It’s worth noting that telcos face challenges when adopting and implementing AI, with skill shortages being the most prominent. For AI, 34% of respondents believe that the lack of skilled labor is the biggest challenge, while for generative AI, this number is even greater (55%). The second biggest challenge is establishing a solid business case with a clear return on investment (ROI). Therefore, 33% of telcos face difficulties in quantifying ROI.

That’s a big chunk of companies that can’t quantify the value of AI. This could be a problem in the continued funding of AI, as telcos have been burned in the past with the promise of “if you build it, they will come.” Being able to measure the value is critically important in understanding where AI is working and not working.

Despite these obstacles, investments in AI are increasing to modernize legacy systems and meet business needs. The report found that 43% of telcos invested over $1 million in AI in 2022, which increased to 52% in 2023. In 2024, 66% of respondents plan to increase their AI budget. There is also a noticeable shift towards larger investments, where 7% of respondents reported investing over $50 million.

Telcos are prioritizing the optimization of customer experience as their top AI investment. This is reflected in the survey responses, where 49% of telcos said they are investing in generative AI for customer service and support. Additionally, telcos are diversifying their AI investments across various use cases, including security, network predictive maintenance, network planning and operations, and field operations. The use of AI in detecting fraud in financial transactions and payments has also increased.

How AI can improve customer service for telcos (no small feat)

The focus on customer experience is long overdue for the telecom industry, as it’s notorious for bad service. In fact, despite making huge investments in its contact center, Comcast is infamous for having the worst service in an industry that’s marred by bad service.

Prior to being an analyst, I was at ITPro, and the one group of companies I never wanted to call was my telco, as I knew I was due for some pain. Generative AI can greatly improve customer interactions as agents are typically put in the painful position of being the integration point for many different systems. Generative AI can do that heavy lifting, putting the agent in a position to be successful.

In 2023, 31% of telcos invested in six or more different AI applications, indicating a more comprehensive range of AI apps used in the industry. Half of the telcos surveyed expressed the desire to move from pilot projects to full-scale implementation of AI. Since AI is becoming a core component of business operations, the focus on concerns like economic uncertainty, infrastructure upgrades, and market differentiation is lessening.

Around 40% of telcos use their own data to train generative AI models internally to improve their current solutions. Additionally, 29% are collaborating with partners to create or personalize models. Regarding deployment preferences, 40% of the respondents favor on-prem, highlighting the significance of low latency and quick outputs for 37% of them.

AI in the cloud or on-premises – or both

As AI becomes more widely adopted by telcos, the choice of platform for running AI is becoming more strategic. Currently, on the one hand, the trend leans towards cloud hosting, with 31% of telcos running most of their AI workloads in the cloud. On the other hand, 25% are running their workloads on-premises. A hybrid configuration – combining cloud and on-premises – is still the most common hosting model, but it’s decreasing in popularity. In 2024, 49% of telcos plan to remain hybrid, 33% will be in the cloud, and 18% will be on-premises.

Partnerships will play a critical role in the adoption of AI in the telecom industry. Engaging with third parties for AI technology is a top priority, with 49% of respondents developing AI solutions for both internal and external users. However, preferences for generative AI are nuanced, with 40% using it to augment existing solutions and 29% opting to build customized models with partners due to data protection regulations.

Looking ahead, AI is expected to become a core component of tech and investment strategies for telcos, enabling the development of next-gen networks, apps, and services. I believe it can help improve customer experience and lower costs. I think it has the potential to create new revenue streams, but this would require telcos to rethink their service approach.

Zeus Kerravala is the founder and principal analyst with ZK Research. Prior to ZK Research, Kerravala spent 10 years as an analyst at Yankee Group.