SK Telecom Wants to Export the Sovereign AI Playbook

Why It Matters

The strategy signals telecoms becoming key AI‑infrastructure providers, reshaping competition in APAC and meeting the massive capital demand projected for AI data centers through 2030.

Key Takeaways

• SK Telecom builds 1 GW AI data center in Ulsan
• Leverages SK Group's telecom, Hynix chips, and AI services
• Targets inference‑focused workloads as AI models mature
• Partners with Anthropic and OpenAI for AI data center projects
• Addresses power challenges via LNG plant and cooling innovations

Pulse Analysis

South Korea’s largest telecom operator is turning its extensive network expertise into a full‑stack AI platform, a move that blurs the line between traditional connectivity services and high‑performance computing. By bundling SK Hynix’s high‑bandwidth memory, real‑estate assets, and AI software capabilities, SK Telecom creates a vertically integrated offering that rivals pure‑play data‑center firms. This sovereign AI model not only promises tighter control over hardware‑software co‑design but also provides a template for other telcos seeking to monetize their infrastructure in the AI era.

The Ulsan campus, slated to deliver one gigawatt of AI‑optimized power, underscores a strategic pivot toward inference workloads, which now dominate AI consumption as models transition from research to production. Partnerships with Anthropic and OpenAI give SK Telecom early access to cutting‑edge models, while its $100 million Anthropic stake signals confidence in the inference market’s revenue potential. To mitigate the notorious power‑supply bottleneck, the company is deploying an LNG‑powered plant and leveraging cold‑energy cooling, ensuring stable, low‑cost electricity for latency‑sensitive AI services.

Across APAC, rivals such as Singtel, NTT, and hyperscalers are racing to secure AI‑ready capacity, but they face similar supply‑chain, regulatory, and energy challenges. SK Telecom’s integrated approach—combining telecom reach, semiconductor backing, and AI expertise—offers a competitive edge that could accelerate its global expansion. As McKinsey projects $5.2 trillion in AI data‑center capex by 2030, operators that can deliver inference‑centric, power‑secure solutions are poised to capture a significant share of the emerging market.