[비즈한국] SK Hynix000660 has confirmed an additional investment of 21.6081 trillion won for the completion of the first fab at its semiconductor cluster in Yongin. Following a board resolution on February 25, the company disclosed a new facility investment for the “construction of the Yongin Semiconductor Cluster Phase 1 Fab and Cleanroom Phases 2-6.” The investment period is set from March 1, 2026, to December 31, 2030.
This disclosure cannot be summarized by a single number. The 21.6 trillion won represents about 29% of the company's consolidated equity as of the end of the most recent fiscal year. Behind the familiar phrase of “responding to AI demand” lies a long-term bet that simultaneously shakes up both the company’s financials and the industry’s cycle.
The key is the timeline, not just the expansion. With this additional investment, SK Hynix has brought its total investment for the Phase 1 fab to approximately 31 trillion won, combining it with the initial investment (approximately 9.4 trillion won) announced in July 2024. At the same time, it has pulled forward the initial cleanroom opening date by three months, from the original May 2027 to February 2027.
21.6 Trillion Won Facility Investment... The Real Impact Depends on the ‘Timing of Equipment Installation’
What should be noted first in this disclosure is the nature of the capital. The 21.6 trillion won announced by SK Hynix is a facility investment focused on the construction of the Phase 1 fab and cleanroom Phases 2-6. Semiconductor fab investments typically see steeper cash outflows during the cleanroom (power, air conditioning, piping, etc.) and equipment installation phases rather than during civil engineering or structural work. The duration, spanning until 2030, serves as a clue for estimating “when the cash will be spent most heavily.”
Therefore, the annual Capital Expenditure (Capex) profile that the market is curious about boils down to when the cleanrooms will “open” and when the equipment will “enter.” Moving the target for the first cleanroom operation to February 2027 is not just a schedule change, but a practical declaration that determines the success or failure of the investment. If the line is not ready by the time supply is needed, “preemptive supply” immediately becomes “late chasing.”
The opposing frame also opens up. The memory industry has historically repeated a cycle where increased demand leads to increased investment, eventually resulting in oversupply and a price crash. Even if AI is creating structural growth, a five-year investment plan is difficult to reverse once a boom ends. Ultimately, there is only one question attached to this investment: Has AI demand solidified into a “trend” rather than a “cycle”? This 21.6 trillion won investment is SK Hynix’s bet that the answer is “yes.”
Beyond the Factory: Power, Partners, and Customers Complete the ‘Cluster’
Another point to watch in the Yongin investment is what “cluster” actually implies. While a fab is just a single factory building, a cluster is completed only when power, water, roads, and the concentration of partner companies all work together. Power, in particular, is the absolute prerequisite for factory operation. Even if factory processes proceed as planned, delays in power grid connections, permits, or utility access will push back the cleanroom opening. Conversely, if infrastructure accelerates, the company’s goal of starting operations in February 2027 becomes more realistic. This means the Yongin investment is a complex project where “corporate investment” and “regional infrastructure” must advance in tandem.
The same logic applies to the partner ecosystem. For a fab to operate, the supply chain for equipment and materials must function stably in close proximity, and base infrastructure like logistics, power, and water must support the concentration of partners. Ultimately, the Yongin investment is not just about building a factory; it is tied to the success or failure of regional industrial policy, including the network of materials, parts, and equipment (Sobu-jang) suppliers.
Finally, there are the customers. It is difficult to increase supply quickly for key AI memory like High Bandwidth Memory (HBM) because it involves complex verification, yields, and packaging. What customers in this market want is not just a price cut, but “receiving what they need at the time they need it.” By setting a firm deadline for the completion of the Yongin Phase 1 fab by 2030, SK Hynix is sending a message to customers that they can incorporate volume for the coming years into their product design and shipment plans. From this perspective, the Yongin investment becomes a prerequisite for “supply chain lock-in.”
However, lock-in is a double-edged sword. Competitors are also increasing investments, and customers have incentives to diversify their supply lines. The moment the pace of supply expansion outstrips the pace of demand growth, the memory price variable will rear its head again.
SK Hynix has emphasized that it has secured investment capacity based on its recent record-breaking performance. Now, the question the market is shifting toward is not “Can they spend the money?” but “When will that money be spent, and how certain is customer demand at that time?” The additional 21.6 trillion won investment in Yongin is SK Hynix’s biggest answer to that question.