The telecommunications market continues to ride technological and operational waves that have drowned sky high cloud ambitions, but Amazon Web Services (AWS) sees ongoing opportunities in advanced radio access network (RAN) technologies like cloud RAN, AI-RAN, and hosted 5G cores to help drive market options.

Ishwar Parulkar, CTO for telecom and edge cloud at AWS, said that despite “announcement and changes in other hyperscalers, we are very committed to the telecom vertical.”

“We believe it is a significant opportunity for us in terms of what we can do with our services and infrastructure, but we also believe the industry is ripe for this transformation,” Parulkar added, later adding that, “we are willing to invest the time and patience for big opportunities to materialize, and we have a lot of patience in that sense.”

That time and investment has already seen AWS become one of the larger cloud-based agents serving the telecom vertical. This includes long-standing work hosting virtualized core network functions and services in support of greenfield and brownfield deployments.

Most recently that work has been tied to cloud RAN and AI-RAN efforts. These tap into AWS’ broad hardware and software infrastructure to help telecom operators more efficiently deploy and manage expensive RAN components.

Parulkar explained that AWS was making progress on recent cloud RAN-linked launches that were announced earlier this year.

“If you look at cloud, it's a multitenant, large-scale, compute storage infrastructure, and there's no reason networks can't be built that way. That was our thesis, and we are kind of proving out the thesis over time,” Parulkar said of this telecom cloud shift.

This thesis includes a value component.

“I think this is an opportunity for telecoms to regain some of the lost ground in the value chain,” Parulkar explained. “When over-the-top players came about, the telco’s value was kind of pushed down the stack in terms of just providing connectivity. And when a technology transformation of this scale happens, this is an opportunity for everybody to play a role. And I think we see this as a huge opportunity.”

Large players are indeed taking advantage of this model.

AT&T, for instance, is amid a multibillion-dollar open and cloud RAN transformation that has a goal of supporting 70% of its wireless traffic through its open RAN platform by the end of 2026. The carrier recently completed a network test that showed the ability to control cloud RAN and traditional RAN components from the same platform.

“The fact that we're building open interconnection at that point of the network means that we can have a competitive process of how those radios get built and how we think about them, and that probably allows us to build it less expensively than we might have five years ago,” AT&T CEO John Stankey said during a recent press briefing.

However, that adoption is still very reliant on a handful of large, leading-edge carriers, a fact Parulkar said is taking time to work through the system.

“It's taking the industry a little while to change it up,” Parulkar said of that pattern. “It's just the scale of the industry and timing of when these transitions happen. These are big investments that telcos make. So it's just the nature of the business.”

AI in the RAN, 5G core in the cloud

AWS is also offering up its AI infrastructure to support RAN deployments. The hyperscaler has worked with operators on incorporating AI into the RAN and was a founding member of the AI-RAN Alliance.

Analysts have pointed to the benefits AI can bring to the deployment and management of cloud-based open RAN networks, which are more complex orchestration challenges due to the disaggregated multivendor ecosystem. The use of AI could help close performance gaps for open RAN architectures compared with legacy RAN models.

Parulkar noted that AI and generative AI can be key to driving RAN efficiency.

“That's a massive opportunity to really build very efficient radio access networks,” Parulkar said. “The problems in this space are very challenging. It is coverage, it's load, it changes with load. There's power management. It's 70% to 80% of the power in the network is consumed in the radio access network. So a tremendous opportunity for AI.”

Parulkar did add that AI use itself is also impacting telecom network traffic patterns, noting that today’s networks were engineered to deal with download-heavy video traffic. This is often seen by wireless telecom operators dedicating more spectrum bandwidth to downlink as opposed to uplink.

“But now we see bi-directional traffic. We see bursty traffic,” Parulkar said. “So the traffic profile is changing, and it'll be interesting to see how telcos take advantage of that to reshape their edge and the network fabric itself to help run these applications efficiently across the network, or move data to compute and compute to data across the network. It's still early, but definitely we're seeing different patterns emerging.”

Analysts are predicting AI-RAN spend could top $10 billion by the end of the decade.

AWS has also found success in hosting 5G cores in both greenfield and brownfield deployments. This includes Comcast’s decision earlier this year to move its 5G core from on-premises to AWS.

This move can provide operators with a more flexible network architecture, which was recently highlighted by EchoStar's decision to ditch Boost Mobile’s own physical 5G network infrastructure in favor of integrating its AWS-based 5G core with AT&T and operate through a mobile virtual network operator (MVNO) model.

Parulkar noted that AWS has had success in past MVNO launches that have used the cloud giant to host those MVNO cores, specifically pointing to work with MVNO Working Group Two.

“They've been pretty successful with MVNOs because MVNOs can just spin up a core, multitenant core and share radio access, or lease radio access from a large operator, and get going, so we're seeing cases of that as well,” Parulkar said.

Private 5G pause

That’s not to say that AWS is not willing to shuffle resources. The hyperscaler earlier this year ditched its AWS Private 5G platform that at launch was targeted at the yet-to-materialize private 5G boom. This comes as market success has so far been limited to specific verticals.

Parulkar noted that the private 5G space “has not materialized as it was expected,” pointing to the continued “bespoke” nature of deployments that doesn’t “lend itself very cleanly to a horizontal platform.”

That scale challenge also continues to be impacted by spectrum issues. Parulkar explained that most U.S. deployments have relied on the quasi-licensed Citizens Broadband Radio Service (CBRS) spectrum band, which is a model that has yet to see similar adoption in other parts of the world.

“There's a lot of larger external factors that have … I would say either delayed or stalled [private 5G adoption,” Parulkar said. “If things change dramatically and if it's a significant market and business, things could be different, things could change.”

CORRECTION: This story has been updated to correct AWS' AI-RAN work telecom operators and its involvement in the founding of the AI-RAN Alliance.