In an unexpected pivot from its traditional telecom roots, SK Telecom has launched the Haein platform, integrating over 1,000 Nvidia Blackwell GPUs. This development, which went live on August 5, 2025, positions SK Telecom as a key player in South Korea's growing AI infrastructure, creating one of the largest GPU clusters in the country.
The Haein platform extends beyond connectivity; it offers GPU-as-a-Service (GPUaaS) and is designed to support South Korea's national AI foundation model program. By managing both training and inference—two of the most demanding processes in artificial intelligence—Haein aims to deliver stable AI capabilities across various sectors.
Strategic Partnerships Fueling Growth
To enhance its capabilities, SK Telecom has partnered with VAST Data to implement what the company calls its “AI Operating System,” specifically tailored for the Nvidia Blackwell infrastructure. This collaboration, announced on August 14, 2025, also includes contributions from Supermicro, focusing on optimizing virtualization and data pipeline performance in multi-tenant environments. Such partnerships help SK Telecom apply advanced technologies to improve its service offerings.
The broader vision of SK Group's AI factory initiative includes deploying more than 50,000 Nvidia GPUs, with SK Telecom playing a central role in this ambitious buildout. The new Haein platform builds upon a previous cluster that utilized Nvidia H100 chips, demonstrating the company's commitment to scaling its AI capabilities.
A Vision for a Sovereign AI Infrastructure
SK Telecom's entry into AI infrastructure reflects the increasing demand for sovereign AI solutions—an aspect that sets it apart from global cloud giants like AWS, Google, and Microsoft. South Korea's government has shown a clear preference for AI infrastructure that remains within national borders, and SK Telecom’s focus on domestic operations aligns perfectly with this strategy.
Being selected as a provider for the national AI foundation model program further validates SK Telecom’s approach, indicating a strong competitive advantage in a space that is increasingly vital for national security and technological sovereignty.
The Decentralized Compute Narrative
The implications of SK Telecom's GPUaaS model extend into the decentralized compute narrative, where projects like Render, Akash, and io.net are working to establish decentralized GPU marketplaces. While these projects aim for a distributed model, SK Telecom's centralized approach, supported by its significant annual revenue of over $16 billion and government backing, highlights a different strategy that may appeal to enterprises seeking reliability and security in their AI deployments.
As of June 2026, SK Telecom and Nvidia unveiled plans to scale this infrastructure to a gigawatt-scale AI Cloud using the Nvidia DSX platform. The first AI factory powered by this initiative is expected to be operational by 2027, targeting applications in AI training, inference, and robotics, among others.
The evolution of SK Telecom from a telecom provider to a leading AI infrastructure player reflects a significant trend in the industry, where traditional companies are embracing AI technologies to stay relevant. As demand for stable AI solutions grows, competition among both centralized and decentralized players in this space will intensify. Whether SK Telecom can maintain its lead in this dynamic environment remains to be seen, but its early investments and strategic partnerships position it well for future growth.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
