SK Telecom transforms into AI-native enterprise with trillion-parameter models and gigawatt-scale infrastructure

At the annual Mobile World Congress in Barcelona, SK Telecom has presented a definitive roadmap to transition from a traditional telecommunications provider into a fully realized AI-native enterprise.[1][2] The strategy, unveiled by CEO Jung Jai-hun during a comprehensive press conference, represents a fundamental restructuring of the company’s operational DNA rather than a mere layering of digital tools over existing services. By rebuilding its core systems, expanding infrastructure to a massive scale, and significantly advancing its proprietary language models, the South Korean operator is positioning itself as a central pillar in the global AI ecosystem.[3] This pivot aims not only to secure the company’s future revenue streams but also to propel South Korea into the ranks of the world’s top three AI superpowers. The move signals a shift away from the commodity connectivity business, often described as the dumb pipe trap, toward a model where intelligence is the primary product and service.[4]

The cornerstone of this transformation is an unprecedented expansion of physical infrastructure designed specifically for artificial intelligence.[5][1] SK Telecom has committed to the construction of hyperscale AI data centers across South Korea with a combined capacity exceeding one gigawatt.[3][5][6][1][7][2] This gigawatt-scale ambition is intended to establish the nation as a primary AI data center hub for the Asian market, attracting global investment and providing the heavy-duty computing power required for next-generation workloads.[2] Central to this plan is a collaborative project with OpenAI to build a specialized AI data center in the country’s southwestern region.[3][6][1][8][7] To manage this massive footprint, the company is leveraging its proprietary virtualization solution, Petasus AI Cloud, which allows for the efficient allocation of GPU resources. Furthermore, the company is tapping into the collective strength of the broader SK Group, coordinating with affiliates like SK Hynix for specialized semiconductor technology and SK ecoplant for advanced energy and cooling solutions.[1] This integrated value chain is designed to optimize the high energy demands and heat signatures of modern AI clusters, which typically struggle within traditional data center environments.

On the software and intelligence front, the operator is significantly raising the stakes in the large language model race. The company’s flagship sovereign AI model, known as A.X K1, currently operates with 519 billion parameters, making it the largest of its kind in South Korea. However, the 2026 roadmap outlines a plan to upgrade this model to more than one trillion parameters by the second half of the year.[3][8] This expansion is paired with a move toward multimodality, enabling the system to process and generate not only text but also voice, image, and video data.[2][1] Unlike the general-purpose models developed by global tech giants, SK Telecom is focusing on industry-specific optimization. By training models on specialized datasets relevant to telecommunications, manufacturing, and public services, the company seeks to provide high-performance solutions for enterprises and government bodies that require localized data sovereignty and enhanced security. This sovereign AI approach is intended to offer a secure alternative for organizations that are restricted from using overseas cloud platforms due to regulatory or national security concerns.[5]

The technical overhaul extends deep into the network architecture and the back-office systems that govern the company's daily operations. SK Telecom is currently rewriting its integrated IT systems, including its billing, sales, and line management platforms, to be natively optimized for AI.[8][9][7][1][3][6] In the past, legacy telco systems were often rigid and siloed, but the new architecture will allow for real-time data analysis to create hyper-personalized service offerings. This includes the ability to automatically design and suggest pricing plans or roaming packages based on an individual customer’s specific usage patterns and behavioral data.[2] Additionally, the network itself is evolving through the deployment of AI-RAN technology and autonomous network operations.[3][9][7] These systems use machine learning to manage wireless quality, optimize traffic flow, and predict equipment failures before they occur, reducing the need for manual intervention and lowering latency across the board. To secure this new intelligent perimeter, the company is implementing a Zero Trust security framework, which utilizes AI-based monitoring to constantly verify every access point and data transfer within the network.

Beyond infrastructure and networking, the company is focusing on a radical redesign of the customer experience and its own internal corporate culture. The existing A-Dot service is being evolved into a comprehensive AI agent capable of managing complex tasks such as call summarization, schedule coordination, and the autonomous execution of follow-up actions.[1][8][2] This agentic approach is designed to provide a single, consistent touchpoint for customers across various digital and physical channels. Internally, the transformation is just as rigorous. SK Telecom has introduced an AX Dashboard to track the adoption and impact of AI across all departments and has established an AI Board to oversee the company’s long-term transformation goals.[3][2] Perhaps most notably, the company has created an internal AI playground that allows employees to build their own no-code AI agents.[3] As of early 2026, more than 2,000 such agents are already active within the company’s marketing, legal, and public relations teams, demonstrating a bottom-up adoption of the technology that complements the high-level strategic directives.

The implications of SK Telecom’s strategy reach far beyond the borders of the South Korean peninsula, offering a potential blueprint for other legacy telecommunications companies worldwide. As growth in traditional mobile subscriptions plateaus globally, the industry is closely watching whether SK Telecom can successfully monetize this shift into AI services and infrastructure. By leading the Global AI Telco Alliance, the company is also working to export these solutions and collaborative models to other international operators, creating a unified front against the dominance of big tech platforms. However, the strategy is not without its risks.[4] Industry analysts have pointed out that the economic viability of running trillion-parameter models on a telco-scale remains unproven, particularly regarding the immense power and cooling costs associated with gigawatt-scale data centers. Nevertheless, the bold announcements at MWC 2026 suggest that SK Telecom views these investments as essential survival measures in an era where connectivity alone is no longer enough to maintain market leadership.

In conclusion, the vision laid out in Barcelona represents a high-stakes bet on the total convergence of telecommunications and artificial intelligence. By committing to a massive infrastructure build-out, the development of massive-scale sovereign models, and a complete modernization of its core IT stack, SK Telecom is attempting to move up the value chain.[1] Success would mean transforming from a utility provider into a central engine of the global intelligence economy. While the costs and technical challenges of such a move are significant, the operator’s comprehensive approach—spanning from silicon and energy to customer-facing agents—positions it as one of the most ambitious players in the current technological landscape. The coming years will determine if this AI-native foundation can deliver the financial returns and customer satisfaction levels required to sustain such a capital-intensive transformation. For now, the company has set a high bar for what it means to be a modern telecommunications company in the age of artificial intelligence.[9]