Wireless Giants: AI Foundational to Next-Gen Networks

WASHINGTON, May 7, 2025 – Nearly every major wireless industry representative speaking at the 2025 CTIA 5G Summit Tuesday delivered a strikingly unified message: The future of wireless will be cloud-native, AI-integrated, and built on Open Radio Access Network technology..

Across a full day of keynotes and panels, leaders from AT&T, Boost Mobile, Ericsson, Intel, Samsung, and T-Mobile outlined a common three-part vision – cloud, AI, and Open RAN – as the foundation of next-generation wireless. But while the vision was shared, execution strategies diverged on scale and timeline.

“We’re able to innovate at the speed of the cloud,” said Even Albertyn, CTO of Boost Mobile, who touted Boost’s network as the world’s largest cloud-native Open RAN deployment operating in the public cloud. “Open RAN isn’t a science experiment. It works.”

Boost Mobile, under DISH/EchoStar, began building its greenfield wireless network in 2019. From the start, it was designed as a cloud-native, fully Open RAN network — meaning all elements are built on open, interoperable standards rather than proprietary vendor solutions.

At AT&T, executives said 70% of their network traffic will run over open interfaces by 2026 – a sweeping modernization effort that includes overhauling nearly every cell site. The company is also emphasizing network programmability, opening software systems to developers and partners to unlock new services and automation.

“We’re exposing that network through APIs to enable innovation at scale,” said Yigal Elbaz, AT&T’s CTO for network services.

T-Mobile pointed to its early deployment of a standalone 5G core and its real-world use of network slicing in stadiums, airports, and emergency services.

“We built this network to deliver personalized experiences on demand,” said Ulf Ewaldsson, T-Mobile’s president of technology. “And now we’re showing how slicing can serve retail, healthcare, and first responders.”

Network slicing allows carriers to create virtual “slices” of a single physical network, each tailored to specific use cases with guaranteed performance and reliability. Paired with cloud-native design, slicing enables real-time responsiveness and dynamic resource allocation, essential for AI applications that depend on low latency, edge processing, and scalable data handling.

Intel’s Caroline Chan framed AI as a catalyst for wireless innovation, especially as workloads shift closer to users.

“AI is migrating from the cloud to the edge – and networks have to be ready,” she said. “Without pervasive connectivity, we can’t modernize factories or scale AI.”

Samsung’s Magnus Ojert emphasized AI’s role in optimizing network efficiency and reducing costs — but said spectrum policy will determine how far that efficiency can scale.

“With AI, we can turn radios on or off based on traffic — Verizon saw a 15% energy reduction,” Ojert said, with an opportunity to save up to 35%. He was referencing results from Samsung’s AI-powered vRAN deployed by Verizon.

Ericsson’s Peter Linder underscored the broader stakes of convergence.

“Cloud, AI, and mobile are converging – and mobile networks have to be ready before the use cases arrive,” he said. “Every AI service will need connectivity. If the network becomes the bottleneck, the whole ecosystem slows down.”

Despite variations in how each carrier and vendor plans to execute, the consensus was clear: AI will both power and depend on smarter, more programmable networks.

Correction: A previous version of this story misspelled the names of AT&T executives Jennifer Robertson and Yigal Elbaz in the photo caption and body text. The story has been corrected.

Correction: A previous version of this story misstated the amount of the energy reduction that Verizon saw. It was 15% (with an opportunity to save up to 35%), not 50%. The story has been corrected.