South Korea's KT Corporation has announced a sweeping 18 trillion won (approximately $13.8 billion) investment strategy, with a significant portion directed toward building 1 gigawatt (GW) of AI data center capacity, as the telecom giant seeks to emerge from a period of management uncertainty and pivot aggressively into artificial intelligence infrastructure.
KT CEO Park Yoon-young unveiled the medium- to long-term roadmap at a press briefing in Seoul on July 6, marking his first major strategic announcement since taking office in March. The plan, which comes 100 days into his tenure, is designed to stabilize the company's core telecommunications business while accelerating investment in high-growth AI and digital infrastructure. "There was a period when KT may have appeared to be experiencing a management vacuum, but employees continued fulfilling their responsibilities," Park said. "I felt that responsibility now rests with management."
The investment is split into two main pillars. KT will allocate 12 trillion won over three years to strengthen its core telecom operations, including network infrastructure, information security, and IT services. Of that, 8 trillion won is earmarked for next-generation network technologies such as 6G wireless, satellite communications, and data center interconnect (DCI). The remaining 4 trillion won will go toward a company-wide security overhaul based on zero-trust principles, a cloud-native transformation, and a doubling of its cybersecurity workforce. Lee Sun-joo, head of KT's Human Resources Division, said the company plans to recruit more than 140 new employees this year, with a particular focus on cybersecurity talent through partnerships with Seoul National University and affiliates like BC Card and K Bank.
The second pillar targets 6 trillion won over five years for AI transformation (AX) infrastructure, with the largest share—5 trillion won—dedicated to building AI data centers (AIDCs) with a combined capacity of 1 GW. Park Hyo-il, head of KT's Strategy Division, emphasized that construction would proceed in stages based on actual customer demand, given the long lead times for land acquisition and facility permitting. KT is evaluating sites in the Seoul metropolitan area based on power availability and permitting conditions, while regional facilities will be developed only after sufficient customer demand is secured. Responding to comparisons with larger investments from SK Group and global tech companies, CEO Park defended the scale of the plan. "We are investing within KT's capabilities and based on actual demand," he said. "This is not based on hypothetical demand but on projects we can clearly execute."
KT also outlined a 1 trillion won investment in submarine cable infrastructure to handle surging data traffic driven by AI adoption, aiming to secure more than 90 terabits per second (Tbps) of additional capacity. Rather than going it alone, the company plans to pursue joint investments with global hyperscale technology companies, which account for the majority of international data traffic. Park acknowledged that substantially expanding submarine cable infrastructure would be difficult for KT to accomplish independently and that partnerships would be the preferred financing model.
In a move that signals a new revenue stream, KT introduced "Token Factory," an AI business designed to manage the billing and consumption of AI tokens. As AI services shift from flat monthly subscriptions to usage-based pricing, enterprises face growing complexity in tracking token consumption across employees and business functions. KT plans to combine its telecom billing expertise with AI token management. "Until now, services such as ChatGPT have largely been offered through monthly subscription fees," CEO Park said. "In the future, AI usage is likely to shift toward token-based metering and pay-as-you-go pricing. The key to solving this challenge is a token gateway." Park Sang-won, head of KT's AX Business Division, said Token Factory will focus on three core principles: optimizing GPUs and NPUs to generate tokens at the lowest cost, routing AI models based on context to minimize token consumption, and optimizing emerging AI agents.
The plan underscores a broader industry trend in which traditional telecom operators are reinventing themselves as AI infrastructure providers. By building massive AIDC capacity and developing token-based billing systems, KT is positioning itself to capture value not just from connectivity, but from the entire AI value chain—from data transport to compute and usage metering. With investment expected to accelerate in the second half of the year and roughly half of the total spending targeted for 2027, KT's roadmap represents one of the most ambitious transformations by a major Asian telecom operator in the AI era.
Source: thelec