AI / LLMMulti-region
← Sticker Wall
JOURNEY · DETAIL
SK Telecom: From South Korea's Largest Mobile Operator to AI Chip and Gigawatt Data Center Transformation
Founded: The company originated from the takeover of the former Korea Mobile Telecom by the SK Group (formerly Sunkyong Group), rather than being a startup founded by individuals. · SK Telecom
JOURNEY
Key Fields
FIELD STAMPSIndustryAI / LLM
RegionMulti-region
ScaleGiant
ChannelOther
Origin
SK Telecom's predecessor was Korea Mobile Telecom, established in 1984. It was privatized and acquired by the SK Group's predecessor, Sunkyong Group, in 1994, and completed the world's first commercial CDMA launch in 1996, rebranding as SK Telecom in 1997. For over two decades, it secured its position as South Korea's largest operator through mobile calls, SMS, and data subscriptions. However, after 2010, voice and data revenue plateaued, and the telecom industry became a low-growth utility. Management concluded that a business model based solely on connectivity had no future and that they must migrate to the upstream value chain of computing power, chips, and AI services, leading to the formal 2022 strategy to transform into an AI company.
Milestones
1984
State-owned origins and privatization Turning point
In 1984, the South Korean government established Korea Mobile Telecom, holding a monopoly on domestic mobile communication licenses. In 1994, a subsidiary of the Sunkyong Group completed its privatization, making it a core asset of the SK Group. In January 1996, SK Telecom achieved the world's first commercial CDMA mobile communication, establishing a technological first-mover advantage and a nationwide user base, subsequently reaching the top market share in South Korea. This phase lasted from 1984 to 1996.
1997
Golden age of mobile business and growth plateau Transition
After rebranding to SK Telecom in 1997, the company enjoyed twenty years of growth driven by 3G/4G network upgrades and smartphone adoption, becoming South Korea's largest mobile operator with a market share consistently near 50%. However, around 2015, as voice revenue declined, data unit prices fell, and regulatory pressure increased, mobile ARPU continued to drop. The company realized the growth engine of pure connectivity had stalled and began exploring side businesses like IoT, media (acquiring broadband TV), and e-commerce (11st), most of which failed to form a second growth curve. This phase lasted from 1997 to 2019.
2020
First bet on AI chips and business spin-off PMF
In 2020, SK Telecom established the AI chip subsidiary Sapeon Korea, launching the X220 inference chip for data centers, aiming to reduce reliance on general-purpose GPUs for operators and cloud data centers using self-developed NPUs. In November 2021, it spun off its investment and semiconductor holding businesses into SK Square, achieving governance separation between the operating company and the investment company, creating capital room for large-scale AI M&A. This phase lasted from 2020 to 2021.
2022
Official announcement of AI transformation and strategic investment Growth
In November 2022, SK Telecom announced three major strategies, declaring the redefinition of fixed-mobile, media, and enterprise businesses with AI, aiming to embed AI into customer service, billing, and sales processes. In August 2023, the company invested $100 million in the US-based large model company Anthropic to jointly develop multilingual models for the telecom industry. Simultaneously, it launched the AI phone assistant 'A.' (A dot) to test consumer willingness to pay for AI services and conversion rates among existing users. This phase lasted from 2022 to 2023.
2024
Chip merger and sovereign AI infrastructure Turning point
In 2024, SK Telecom pushed for the merger of Sapeon with the South Korean AI chip startup Rebellions to form the largest AI chip company in Korea, benchmarking against NVIDIA's inference market. In November 2024, it announced an investment of 100 billion KRW starting in 2025 to package Rebellions' NPUs, SK Hynix's HBM memory, and self-built AI data center solutions into a 'Korean-style sovereign AI system.' However, valuation disagreements and talent integration issues delayed merger negotiations, exposing friction between startup culture and state-owned enterprise processes. This phase lasted from 2024 to 2025.
2025
AI business becomes a financial growth engine Growth
Q1 2025 financial reports showed steady year-on-year revenue growth for both AI Data Center (AIDC) and AI Transformation (AIX) businesses, becoming high-growth segments for the group. In March 2026, the company announced a joint investment of up to 100 trillion KRW with global tech partners to expand the Ulsan AI data center under construction to a 1-gigawatt scale—the largest in Asia—while developing trillion-parameter models. Internally, it uses an AX dashboard to monitor AI usage across departments, with over 2,000 AI agents currently in operation. This phase lasted from 2025 to 2026.
Turning Points
- 1996: The world's first commercial CDMA launch secured its ticket to the mobile communication era.
- 2021: The spin-off of SK Square freed the operating entity to pursue AI M&A.
- 2023: A $100 million investment in Anthropic tied the company to the global large model race.
- 2024: The merger of Sapeon and Rebellions bet on sovereign self-developed AI chips.
- 2026: The announcement of the 1-gigawatt Ulsan data center and 100 trillion KRW investment pushed all chips into AI infrastructure.
Failures & Pitfalls
- Side businesses like media and e-commerce (11st) failed to gain traction for years and could not offset the decline in connectivity revenue.
- Consumer-side conversion for the AI phone assistant 'A.' was lower than expected, with slow penetration of consumer AI subscriptions in South Korea.
- The merger between Sapeon and Rebellions was repeatedly delayed due to valuation disagreements and organizational culture conflicts.
- Early data center ROI was under pressure, leading capital markets to question the capital efficiency of telecom operators entering the computing power business.
关键成功要素
- Using cash flow from the license and network monopoly period to fund AI infrastructure.
- A group-wide synergy strategy integrating chips (Rebellions), memory (SK Hynix HBM), and data centers.
- Partnering with global players like Anthropic to gain large model technology and international expansion channels.
- Using AX dashboards and AI committees to turn transformation into a quantifiable organizational engineering project.
- Betting on the 'sovereign AI' narrative to gain support from the government and local enterprise clients.
Lessons
- The core of telecom operator transformation is selling high-value computing power and model services, not just connectivity.
- Spinning off investment and operating entities can resolve capital operation constraints during transformation.
- Self-developed chips must be built in sync with software ecosystems and customer scenarios, or they risk becoming laboratory products.
- Organizational-level implementation of thousands of internal AI agents is a better test of transformation authenticity than isolated AI applications.
- The biggest risk for giants is diversifying into too many side businesses; focusing on one main line (AI infrastructure) is the only way to achieve a second growth curve.
Core Data
- 2026 Ulsan Data Center planned capacity：1GW (Targeting the largest scale in Asia) (Public data, independent verification not performed)
- Ulsan Data Center joint investment scale：Up to 100 trillion KRW (Public data, independent verification not performed)
- 2023 investment in Anthropic：$100 million (Public data, independent verification not performed)
- Sovereign AI initiative investment (from 2025)：100 billion KRW (Public data, independent verification not performed)
- Number of internal AI agents running：Over 2,000 (Public data, independent verification not performed)
- Target large model parameter scale：1 trillion (Public data, independent verification not performed)
Competitors / Peers
In South Korea, it competes with KT and LG U+: KT is also betting on AI customer service and data centers, while LG U+ focuses on enterprise AI small models; all three see AI as a way to escape commoditization. Globally, it benchmarks against SoftBank and NTT, which have integrated computing power into their strategies—SoftBank positions itself through investments in Arm and the OpenAI ecosystem, while the NTT Data Center Group is already among the top three globally. The deeper, implicit rivals are cloud giants like Amazon AWS and Microsoft Azure; SK Telecom is attempting to carve out a middle ground between them and South Korean government/enterprise clients using a local sovereign AI narrative and 1-gigawatt self-built data centers.
Sources
- https://www.prnewswire.com/news-releases/sk-telecom-ceo-unveils-ai-native-strategy-at-mwc26-driving-koreas-leap-in-ai-innovation-302700470.html
- https://www.sohu.com/a/992189413_122036485
- https://ai.zhiding.cn/2026/0303/3180031.shtml
- https://www.winzheng.com/article/sk-telecom-ai-core-rebuild-mwc-2026
- https://www.toutiao.com/article/7613041635307291186