Telecommunications operators are finally beginning to reap the rewards for their efforts in artificial intelligence, according to chipmaking behemoth Nvidia. According to its fresh State of AI in Telecommunications report, which surveyed 450 telecoms professionals across the world, over 80% of telcos said that AI is helping increase their company’s annual revenue. “There’s a lot […]
Telecommunications operators are finally beginning to reap the rewards for their efforts in artificial intelligence, according to chipmaking behemoth Nvidia.
According to its fresh State of AI in Telecommunications report, which surveyed 450 telecoms professionals across the world, over 80% of telcos said that AI is helping increase their company’s annual revenue.
“There’s a lot of talk about whether AI investments are truly paying off, but according to the survey, telcos are already seeing significant returns,” Chris Penrose, global head of business development for telco at Nvidia, tells TI.
77% of telcos saw improvements in operational costs, and over a fifth reported a return on investment greater than 10%.
“It’s clear that AI is driving tangible results. The biggest change we’re seeing now is that we’re moving out of pilot projects and into full-scale production,” says Penrose, “which is where the real benefits are.”
80% of the respondents believe that AI is “crucial” for their company’s future success, with two-thirds planning to increase spending on AI infrastructure in 2025.
Ahead of Mobile World Congress, in Barcelona, TI speaks to former AT&T executive Penrose on how telcos – and their enterprise customers – can see the benefits of AI adoption.
Where is AI making an impact in telecoms?
Among the multiple use cases that are emerging for generative AI, most companies are seeing its biggest benefit through improved employee efficiency, at 65%.
Over half (54%) use it for customer service and support, and network operations and management come in at third (45%).
The plan is for telcos to not only use it internally but also continue to offer GenAI to their customers, with 84% planning to offer solutions and 52% to provide it as a software-as-a-service.
“Historically, maybe telcos weren’t the first to adopt new technologies, but generative AI presents an opportunity in two big ways,” says Penrose.
He explains that alongside using it to drive employee efficiency and improve customer experience, telcos are looking to gain revenue from more than just 5G.
“Despite investments in 5G, telcos haven’t seen a huge revenue increase from these investments. They haven’t been able to charge significantly more for 5G compared to 4G.
“This creates a real pressure to find new revenue streams.”
He elaborates that GenAI and AI, in general, could provide this opportunity, particularly with the rise of sovereign AI.
Sovereign AI refers to a nation’s capabilities to produce artificial intelligence using its own infrastructure, data, workforce and business networks, according to Nvidia.
“Telcos are in a unique position to support this, given their infrastructure, trust with governments, and access to resources like space and power.”
With many telcos partially or fully government-owned in some countries, they have the benefit of expertise to build large-scale infrastructure, which is particularly important for scaling AI, he explains.
“Sovereign AI offers telcos a chance to expand beyond just providing connectivity and tap into new revenue opportunities by becoming key partners in developing national AI infrastructure.”
A multi-approach in AI development
The report also highlights a trend toward utilising multiple approaches to AI development. In-house AI solutions grew from just 6% in 2023 to 37% in 2024.
“Telcos have often outsourced much of their technology around platforms and day-to-day operations,” says Penrose. “However, we’re seeing more and more telcos establishing AI Centres of Excellence within their companies.”
With telcos handling massive data sets, Penrose believes they can leverage this data to create customised AI models that will not only enhance their own operations but could also differentiate their services in the market.
“I think we’re seeing telcos realise the importance of not outsourcing intelligence around data. They’re leaning into developing more of this expertise internally.”
That said, they’re not doing it all alone; 43% of respondents said that they will continue to engage with third parties to codevelop AI solutions.
“Telcos can use off-the-shelf, high-performance solutions for aspects of their operations that aren’t necessarily a point of differentiation, but that help them run better.
For areas where they want to differentiate, they’re focusing on building their own solutions or working with partners to tailor them.”
Penrose notes that they’re seeing a balance and telcos embracing both approaches, building in-house and using trusted partners.
The help is needed too, since 43% of the respondents said that the need for data scientists, engineers, architects, and developers has been a key obstacle to AI adoption at scale – up from 24% the year before.
Chris Penrose, global head of business development for telco at Nvidia
“That’s part of the reason why many telcos are investing in AI Centres of Excellence, so they can build the infrastructure to test, learn, and develop these technologies.”
“By investing in their own people and processes, they can really start to apply AI effectively.”
“At the same time, there are many trusted partners that can help them on this journey.”
For example, Penrose points to Amdocs, a technology company specifically providing telecoms solutions, who came to Nvidia to develop an AI-driven billing assistant to offer to their customers.
“Now, if a telco wants to deploy an AI-driven billing assistant, they can access that through Amdocs.
“This is a practical solution because around 13% of customer service calls are related to billing inquiries,” says Penrose.
According to Penrose, the next phase is about helping businesses apply AI to their operations.
“Telcos already have strong relationships with businesses of all sizes, especially small and medium-sized ones.
By offering platforms that provide access to AI hardware and software, telcos can help these businesses adopt AI.”
From there, telcos can move into offering more AI-driven services. For example, if a telco creates an AI assistant for internal use, they could potentially offer these solutions to their business customers as well.
The next phase
For operational efficiency within the telcos, firms are focusing on running AI and the radio access network (RAN) on the same infrastructure.
“Telcos spend around $300 billion a year on infrastructure, but they build for peak usage. So, when the network is fully utilised at 5pm but underutilised at midnight, could we repurpose that unused capacity for AI tasks?” asks Penrose.
“This would provide huge economic benefits.”
He also points to using AI to help support the increasing amount of AI traffic, like devices and sensors talking to each other and generating data, while ensuring optimal service, low latency, and time-sensitive outcomes.
According to the survey, more than 50% of telcos expressed interest in sharing network capacity with AI workloads, along with applying AI to improve network performance and efficiency.
“As we move towards 6G, we’ll see AI become fully integrated, benefiting telcos’ internal operations and monetising next-gen services,” which may make it fair to call 6G and AI-native technology.
Next up, a shift towards the new hot technology: agentic AI.
“Where multiple AI systems from different companies work together. This is a key consideration for telcos,” says Penrose.
He sees telcos possibly using AI from different sources, with the challenge of getting the systems to interact with the telco’s internal AI to create better solutions for all.
“Stitching these AI systems together into a cohesive solution is a big trend to watch for the future.”
