Verizon, Telia execs paint the GenAI big picture

Rainer Deutschmann from Telia, Srini Kalapala from Verizon and Lan Guan from Accenture in a panel discussion at Mobile World Congress 2024 in Barcelona - 26 February 2024.

Rainer Deutschmann from Telia, Srini Kalapala from Verizon and Lan Guan from Accenture in a panel discussion at Mobile World Congress 2024 in Barcelona - 26 February 2024.

  • AI was the dominating topic at this year’s Mobile World Congress in Barcelona
  • High-level executives from Verizon and Telia outlined how telcos can harness the power of generative AI (GenAI)
  • Linking GenAI to a solid data foundation and value assessment when prioritising use cases will be key

BARCELONA – MWC24 – Executives from Verizon and Telia stressed the importance of organising data and identifying AI use cases that should be prioritised as telcos seek the best way to unleash new opportunities in the generative AI (GenAI) era.

These were the main takeaways from the ‘Harnessing GenAI at telco scale’ panel discussion held in Barcelona earlier this week.

Srini Kalapala, SVP and chief product development officer at Verizon (pictured above, middle), noted that the telco sector will play a key role in the advancement of generative AI (GenAI), because its use and development will require data transfers at scale, a process enabled by telecom network operators.

In his recommendations to telcos, he suggested they need to optimise the organisation of their data assets. “One of the challenges we all face in the telco world is that we still have network, IT, marketing, different islands of data… We’ve got to bring all of the data together to deliver the right outcomes to the customers.”

Rainer Deutschmann, SVP and group COO at Telia (pictured above, far left), added that to scale GenAI, telcos need to leverage their foundation, which lies in “simple things” like having a common data lake, having the sources integrated into the lake and “having the data products that you can use for either the customer, for the products themselves or for the internal operation.

“Leveraging the foundation is key now and…  I would always want to link GenAI into a solid data foundation. And the data foundation obviously relates to the technology, but also relates to the people. It’s a broader transformation to drive,” said Deutschmann.

It is “a lot better” to build the foundation first, he added, before unlocking the power of GenAI than “trying to do the GenAI as a leapfrog and all of a sudden you realise that foundation is missing – so it has to go hand in hand.”

Another important aspect to consider is for telcos to look for the value. “[Just] because I have so many opportunities, it doesn't mean that I should do all of these things at the same time – and I cannot. And I don’t want to just invest into one of the areas, forgetting an area that may have more value for me,” he maintained.

Telia’s group COO gave the example of what he called a “value board” where the Swedish operator tests through a “grassroots/bottom-up approach” and looks at the feasibility and the value that various use cases have to the company, before selecting “very few that are scaled to production”.

“For me, it’s a key question – is it paying back? And if I do so, should I do this for my entire workforce or for the entire set of developers, or do I just have more choice to bring it to those people and those areas where I get the most bang for the buck,” he asked.

The introduction of GenAI should also not be seen as just one system: Deutschmann believes there will be a lot more fit-for-purpose solutions than we currently have.

“And then there will be what you can call a ‘pot of pots’, so you will [have an] orchestrated set of systems that work together” with “orchestration on top of it”, noted the Telia executive.

Finally, Verizon’s Kalapala acknowledged that GenAI is still at a very early stage, noting that its wide adoption will be possible once inference costs are driven down – which means allowing for inference to be run where it is “more economical”.

“You don’t need to send all the data” back to a centralised platform all the time as that would not be efficient and would be impacted by latency, he explained, adding that research into distributed large language models (LLMs) shows that end user devices and distributed IT systems will play a key role in bringing GenAI processing to the edge of the network.

- Yanitsa Boyadzhieva, Deputy Editor, TelecomTV

Email Newsletters

Sign up to receive TelecomTV's top news and videos, plus exclusive subscriber-only content direct to your inbox.