"Even in the AI Era, Investment Is Still Assessed in a 3G-Style Manner... Government Must Change Telecom Investment Standards"

[Financial News] Calls have emerged for the government to change its standards for assessing telecom operators' investments, which focus on tangible facilities such as base stations and cables. Software development using artificial intelligence (AI) to improve telecom quality and cloud usage costs are not adequately reflected in existing capital expenditure statistics, making it difficult to accurately assess actual investment.
Cho Kyoung-tae, a People Power Party lawmaker representing Saha-gu Eul in Busan and a member of the Science, ICT, Broadcasting, and Communications Committee of the National Assembly, said he plans to call on the government during this year's National Assembly audit to introduce a supplementary indicator that separately tracks telecom operators' AI investments and to expand tax credits for AI intangible assets.
According to Cho, the three telecom operators' nominal capital expenditures fell from about 9.6 trillion won in 2019, the first year of 5G commercialization, to about 6 trillion won last year. They also declined by about 7% year on year in the first half of this year. This has repeatedly fueled criticism that telecom operators are neglecting network investment while focusing on AI businesses. However, Cho pointed out that capital expenditure alone cannot explain the industry's changing investment structure.
Telecom network upgrades are expanding beyond adding equipment to improving operational efficiency through AI. Citing overseas examples, Cho said Australia's Optus increased base station transmission speeds by more than 20% using AI, without adding equipment or spectrum, while Japan's SoftBank Group improved 5G upload speeds by about 30%. Germany's Deutsche Telekom reduced outage response times from about an hour to a few minutes.
However, these costs may be excluded from capital expenditure statistics depending on accounting practices. AI software development costs may be recorded as research expenses, while the cost of renting GPU servers through the cloud may be treated as a usage fee. Equity investments in AI companies are also recorded separately from capital expenditures. As a result, an assessment focused on the number of base stations and investment in tangible facilities is unlikely to capture these changes.
Cho proposed maintaining the existing capital expenditure indicator while regularly compiling and publishing data on software and data investment, AI model training and operating costs, cloud usage fees, AI research and development (R&D), and equity investments in domestic and overseas AI companies. He called on the Ministry of Science and ICT to include such an "AI investment supplementary indicator system" in a comprehensive plan by the end of the year.
He also cited overseas investment indicators as examples. Verizon Communications announces capital expenditures that include software, while France's Orange discloses its own investment indicator covering intangible assets. The United Kingdom's energy regulator Ofgem operates a total expenditure model that combines capital expenditures and operating costs.
Cho also plans to call for expanded tax support. Under the Restriction of Special Taxation Act, he wants the government to consider broadening the scope of intangible assets and new growth and source technologies eligible for integrated investment tax credits, allowing AI software and data investments used to upgrade telecom networks to qualify for deductions. Cho urged consultations between the Ministry of Science and ICT and the Ministry of Economy and Finance and said he would consider introducing related legislation after the National Assembly audit.
Cho said, "Wrong assessment standards lead to wrong numbers, and wrong numbers lead to wrong policies. If we aim to become an AI powerhouse, we must establish investment standards befitting an AI powerhouse."
[email protected] Yoon Hong-jip Reporter