The GenAI hype cycle is in full swing, but AWS’ CTO for telecom and edge cloud Ishwar Parulkar predicted 2024 will be the year of the RAN as operators continue leveraging the cloud to transform their businesses and monetize 5G. First item on their checklist? Finishing work they’ve already started to cloudify their networks.
While many network functions like IMS, OSS and BSS have already moved to the cloud in recent years, Parulkar said there’s still one big puzzle piece missing: the radio access network (RAN). He noted that’s a problem because in order to achieve the full potential of new 5G tools like network slicing, operators need an open and end-to-end cloud-native network. His words echoed a comment made by former Rakuten Symphony North America CEO Azita Arvani at Silverlinings’ recent Cloud Executive Summit.
Part of the problem has been overcoming sunk costs in legacy equipment, long refresh cycles and doubts about the cloud’s reliability. Executing internal cultural changes necessary to accommodate a next gen network has also been a challenge for operators. But Parulkar said the tide is turning.
“The conversation has changed dramatically,” he said. “Now it’s not about why cloud, it’s about how and where.”
He added there’s also recently been “a lot of activity looking at open RAN and this is the year we’ll really get a solid direction around how and where open RAN will add value to operators,” he said. “Once you have a very dynamic software-based, container-based network infrastructure running on the cloud then you can start really creating these slices.”
API acceleration
Parulkar flagged network APIs as another 5G promise that could move closer to fruition in 2024. In theory, 5G allows operators to expose their network capabilities via APIs, which in turn allows developers to create new and exciting applications that capitalize on next-gen network tools. But, like slicing, this has yet to happen on a large scale.
This is partially because it has taken some time for operators to come together to standardize network APIs so they work across operator networks. Only eight standardized APIs were available as of the middle of 2023 via Project CAMARA, but by the end of the year that figure had increased to 20.
The other issue has been getting the APIs in front of developers. And that, Parulkar said, is where hyperscalers can help.
“We have the largest community of developers, from startups to large companies, from financial institutes to governments to retail, you name it. So, we bring that community. It’s difficult for telcos to just offer APIs and expect developers to come and use them,” he said.
Shirin Esfandiari, product marketing director at Oracle, similarly tipped operators to focus more on APIs and forging strategic alliances on that front in 2024. However, she said there are still questions to be answered.
“The industry will continue to grapple with key questions, including what the API monetization models should be and whether or not a neutral third-party host is essential to scaling API adoption across the industry,” she said.
Gearing up with GenAI
How could we end a prediction piece without a mention of what 2024 holds for telcos on the generative artificial intelligence (GenAI) front?
Parulkar said over the past year operators have developed a better understanding of what GenAI can – and, more importantly, can’t – do. Now that the hype cycle is winding down, they’re looking to settle on specific use cases to push out into the wild.
He tipped GenAI to be used for streamlining internal processes, enhancing the customer experience, improving productivity and aiding network operations. For instance, GenAI can be used to help with code development, RFP generation and identifying revenue leakage, he said.
Esfandiari added operators are expected to begin mapping out roadmaps to implementing autonomous networks [AN] that rely on AI and ML technology.
“We foresee more CSPs setting public targets with timeframes for AN maturity withing their organizations with many seeking Level 3 compliance (policy based autonomy and conditional closed-loop control) and some Level 4 compliance (intent driven closed-loop control) as well as seeking collaborative partners with whom to engage in the journey,” she concluded.
It does seem like there may be a bit of a disconnect between what hyperscalers expect and what operators are actually feeling though. During Silverlinings Cloud Executive Summit in December, AT&T's Chris Hristov noted it's actually proving a bit difficult for companies to prioritize AI use cases.