Gambling on the Nation's Fate? South Korea's Trillion-Dollar Bet on Semiconductors, While the Bank for International Settlements Warns of an AI Bubble

TL;DR

· South Korean President Lee Jae-myung will preside over the "Three Mega Projects," elevating semiconductors, AI data centers, and physical AI to a national-level investment direction.

· BIS warns that the AI capital expenditures of the five cloud giants have exceeded profits and free cash flow, and financing pressure on the demand side is beginning to be priced in.

· Related targets: Samsung Electronics, SK hynix, NVIDIA, Microsoft, Google, Amazon, Meta, Oracle, Power Equipment, and Copper Industry Chain.

According to Reuters on June 29, South Korean President Lee Jae-myung will preside over an industry unveiling event, announcing the three major projects covering semiconductors, AI data centers, and physical AI. South Korean media estimates that Samsung and SK's related long-term investment may exceed 1,000 trillion Korean won.

This news places the AI hardware cycle in a new position. The supply side needs national coordination of land, electricity, water resources, and regional layout, while the demand side is being questioned by macro institutions about cash flow sustainability.

Almost simultaneously, in its 2026 Annual Economic Report, the BIS reminded that the AI-related capital expenditures of the five hyperscalers from 2025 to 2026 will exceed $1 trillion, commitments have already exceeded profits and free cash flow, prompting some companies to finance through debt issuance.

What South Korea sees is a window to seize the AI hardware gateway, while the BIS sees U.S. cloud giants continuing to leverage up for AI platform position. Investors need to assess not whether there is AI demand, but whether the demand can continue to be supported by revenue, profit, and cash flow.

South Korea's Focus is on the Hardware Gateway

South Korea's move to elevate the project to the presidential level is essentially a competition for the most irreplaceable hardware gateway in the AI industry chain.

Regular investors can understand HBM (High Bandwidth Memory) as a high-speed memory stack next to an AI chip. AI training and inference require not only NVIDIA GPUs but also continuous memory to feed data in. SK hynix leads in HBM, with Samsung also in pursuit.

The project also covers AI data centers and physical AI. Physical AI involves putting models into robots, devices, and factory systems to allow AI to perform operations in the real world. The demand for it may come from manufacturing, logistics, robotics, and edge devices.

According to reports, South Korean policy advisor Kim Yong-beom believes that the demand for AI is growing faster than expected. South Korea cannot simply allow companies to expand production independently; instead, it must incorporate electricity, land, water resources, and regional layout into its planning. Some discussions also point to areas outside the capital region, such as Gwangju and Jeolla, where infrastructure constraints have become a bottleneck for expansion.

The market implication of this round of South Korean investment is not just an increase in semiconductor spending but an attempt to transform the expansion of Samsung and SK's enterprises into a long-term position for South Korea in the AI hardware supply chain.

BIS Focuses on Demand-Side Cash Flow

The BIS's warning does not directly label AI as a bubble; instead, it points to a gap between capital expenditure and cash flow.

Capital expenditure is the money that companies spend in advance for future growth, such as building data centers, buying GPUs, and expanding power systems. Free cash flow is the money that companies can use freely after deducting necessary expenses. The BIS cautions that the AI capital expenditure of the five major cloud giants has already exceeded profits and free cash flow, and their expansion is increasingly relying on financing arrangements.

This aligns perfectly with South Korea's supply-side bet. South Korea is expanding in HBM, advanced packaging, data centers, and robotics hardware, while American cloud giants are purchasing AI servers, chips, power, and data centers. The investment confidence of the former comes from the order expectations of the latter.

The BIS also presents a more cautious scenario: under competitive pressure, companies may overcommit to investment in order not to lose their position in the AI platform, even if the overall industry's net economic surplus eventually turns negative. Similar processes have occurred during the railway and internet bubble periods, where the technology direction was viable long-term, but at some point, capital allocation became overheated.

This is also the aspect that the market needs to distinguish when the stock prices of Samsung and SK hynix briefly fell on the day following the related reports. While public reports attributed the day's decline more to a global tech stock consolidation after a rise, rather than a single policy factor, the stock prices still indicate a question: when massive expansion meets return validation, supply chain leaders will also be reassessed for depreciation, utilization rates, and order visibility.

Nationalizing Supply to Amplify Demand Validation

South Korea's actions have strengthened the AI hardware cycle and increased its reliance on the capital expenditure of American cloud giants.

In an optimistic scenario, American cloud giants continue to increase AI spending, and enterprise AI services, cloud revenue, and inference demand gradually cover chip, depreciation, power, and interest costs. South Korea's early layout in HBM, advanced processes, packaging, and data centers will then create a national-level industrial moat.

In a more cautious scenario, problems will propagate back along the same chain. If cloud giants find that AI revenue growth cannot cover all costs, they may delay data center construction, reduce server orders, or even renegotiate supply chain prices. The capacity, parks, and power investments that South Korea expanded early on will face pressure on utilization and return on investment.

One of the most easily misunderstood aspects here is the investment amount. In Korean media reports, there are different expressions regarding Samsung's and SK's long-term investments, such as "over 100 trillion KRW" and "a total of about 200 trillion KRW." A more cautious interpretation is that this is a high-level plan or estimate for the next decade, not equivalent to a fully executed government commitment, nor does it mean that the orders will be smoothly absorbed every year.

The South Korean government is aiming to enhance the country's industrial position, while U.S. cloud giants are focused on AI business returns. The former can be promoted through policies, land, and infrastructure development, while the latter ultimately relies on revenue, profit, and cash flow validation.

Power and Utilization Rate Determine Valuation Flexibility

What may be initially validated in this AI hardware cycle is not the grand narrative of productivity but rather a few more robust variables: whether the power supply is in place, if factories are constructed on time, if HBM and server orders are continuous, and if cloud giants can still support capital expenditure with operating cash flow.

Power supply is particularly critical. AI data centers have continuous demands on the power grid, transformers, copper, and cooling systems. If the regional data centers and semiconductor clusters planned in South Korea advance simultaneously, power equipment and bulk commodities will benefit. If grid expansion lags behind project schedules, investment commitments will initially turn into cost pressures.

Another variable is post-2027 utilization rates. The current market is willing to assign a higher valuation to the AI supply chain because the default training and inference demand will continue to grow. Once cloud giants' capital expenditures shift from acceleration to deceleration, the first changes may not be technical assessments but rather order visibility, inventory cycles, and depreciation pressures.

This national-level bet by South Korea is more like an amplifier. In the realization of AI returns, it will amplify Samsung's and SK hynix's positions in the global supply chain. If the returns fall short of expectations, it will also amplify the speed of the capital expenditure backlash. For investors, the most concrete validation point ahead is whether U.S. cloud giants' cash flow can continue to catch up with the AI spending they have already committed to.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia