Ericsson named sole global tech partner in SK Telecom-led AI-RAN pilot

Ericsson set to deploy AI-RAN at KG Mobility under South Korea’s AI Highway plan

In sum – what we know:

• Sole global partner – Ericsson is the only international tech partner in SK Telecom’s consortium for South Korea’s government-backed Hyper-AI Network demonstration project.
• The EIAP play – Its contribution centers on the Intelligent Automation Platform, AI-powered rApps, and open network management tools for autonomous network control.
• A divergent bet – Ericsson and SKT emphasize telco-grade RAN and orchestration, contrasting with NVIDIA’s GPU-centric AI-RAN Alliance approach.

Ericsson has been named the sole global technology partner in an SK Telecom-led consortium for South Korea’s Hyper-AI Network Infrastructure Demonstration Project, a government-backed effort spearheaded by the Ministry of Science and ICT (MSIT) and the National Information Society Agency (NIA). 

The demonstration project is a cornerstone of South Korea’s “AI Highway” plan, which aims to establish the world’s first nationwide AI infrastructure by 2030. The plan comes with explicit goals attached, including 6G leadership and the buildout of AI-native networks. For Ericsson, landing the sole international partner role in one of the two flagship pilots is a meaningful foothold in a market where domestic vendors have historically had the home-field advantage.

Notably, SK Telecom isn’t running this alone at the national level. KT, another major South Korea telecom, is leading a parallel AI-RAN pilot under the same initiative — a deliberate move by the government to keep the competitive landscape balanced while both operators push the country’s physical AI agenda forward.

Ericsson’s technology

Ericsson’s contribution centers on its Intelligent Automation Platform (EIAP), a set of AI-powered rApps, and open network management tools. The pitch is autonomous network control — real-time optimization and resource orchestration across industrial settings, with the network adjusting itself rather than waiting for a human operator to intervene. Whether that holds up in a live industrial deployment is precisely what the pilot is meant to find out.

The technical architecture combines 5G Standalone (SA) capabilities with AI-driven network automation and distributed AI computing. The target workloads fall under what the consortium calls “physical AI,” which covers autonomous transport, humanoid robotics, and industrial safety monitoring. These are demanding use cases. A patrol robot at a refinery or an autonomous vehicle in a logistics yard can’t tolerate the kind of latency hiccups that a consumer streaming video barely notices, which is why the network automation layer is arguably the most important piece of the whole project.

The collaboration isn’t starting from scratch, either. It extends an existing MoU signed between Ericsson and SK Telecom in March 2026, which targets joint R&D, trials, and potential commercialization of AI-RAN and 6G technologies through 2031. This project effectively gives that agreement a government-funded proving ground.

It’s also worth noting how this approach differs from the other major AI-RAN effort in the industry. NVIDIA’s AI-RAN Alliance prioritizes GPU-centric AI training and inference, essentially treating the radio network as another place to put accelerated compute. The SKT and Ericsson approach flips the emphasis toward telco-grade RAN and network orchestration — the network itself as the intelligent layer, rather than a host for GPUs. Neither strategy has been proven at scale yet, and the two aren’t necessarily mutually exclusive. But they represent genuinely divergent bets on what an AI-integrated network should look like.

Timeline

The consortium aims to build and validate its AI-RAN pilot networks by 2027, split across two phases with different vendors, sites, and use cases.

Phase 1 runs from 2026 to 2027 in Incheon and Pangyo, where the consortium will test AI-RAN for patrol robots at an SK Group-owned refinery. Interestingly, this phase will use Samsung Electronics equipment rather than Ericsson’s — a reminder that even as Ericsson holds the sole global partner title, the domestic incumbent isn’t being sidelined.

Phase 2 is where Ericsson’s gear takes center stage. The consortium will deploy Ericsson’s AI-RAN at KG Mobility’s Pyeongtaek plant to test autonomous vehicles in logistics operations, putting the company’s automation stack up against one of the more unforgiving physical AI scenarios on the roadmap.

Throughout the project, SK Telecom will simultaneously evaluate equipment from Samsung, Nokia, and HFR. That’s a sensible hedge for an operator that will eventually need to make long-term procurement decisions, and it means Ericsson’s position isn’t exclusive in any practical sense. Still, Ericsson remains uniquely positioned as the core international partner for end-to-end RAN integration and automation — the vendor responsible for making the whole system work together, rather than supplying one piece of it. If the pilots deliver, that’s the role most likely to translate into commercial deployments when the AI Highway plan moves past the demonstration stage.