Jensen Huang and Nvidia [Photo=Yonhap News]

 The theme of this week's trend is "Earnings are innocent, lifespan is guilty."

Last week, Stock Radar presented the following as key trends:

△ The memory sector rally ignited by Micron Technology,

△ The software diffusion of AI infrastructure,

△ The resurgence of Korean semiconductors.

The question was, "Can the AI rally push Korean semiconductors back to the center of the index?"

The answer in the early part of this week was positive.

Dell Technologies surged on the back of soaring sales of AI-optimized servers and an upward revision of its guidance.

This demonstrated that AI investment is not just about Nvidia, but is spreading across servers, storage, data centers, power infrastructure, and software.

Nvidia's unveiling of AI chips for personal computers (PCs) followed the same trend. AI is moving beyond data centers and into the PC market.

However, in the latter half of the week, the market posed a different question.

Broadcom was not a company with faltering earnings. Micron is also a company that benefits from the AI and memory cycles. Nevertheless, its stock price fluctuated significantly.

What the market observed was not immediate earnings, but the lifespan of the infrastructure assets supporting those earnings.

This week, Stock Radar has one question:

“AI infrastructure investment continues, but how long will the economic lifespan of those infrastructure assets last?”

The faster the performance of AI chips improves, the faster the value of existing equipment depreciates.

Even graphics processing unit (GPU) infrastructure, built by data centers at trillions of won, could see its leasing rates and collateral values shaken by the next generation of chips.

This is not a matter of accounting depreciation, but a question of how long that equipment can generate revenue in the actual market.

At this point, the nature of semiconductor stock prices changes.

Until now, the market bought semiconductors based on the single sentence, "AI demand is increasing." However, going forward, it is likely that investors will consider questions such as, "Which semiconductors can sustain increased demand for a longer period?", "Which companies can afford the cost of generational replacement?", and "Who can absorb the investment burden with their own capital and cash flow?"

This question is also crucial for Korean semiconductors.

Memory demand is a direct beneficiary of increased investment in AI servers and data centers. However, the faster the AI infrastructure investment cycle, the faster the debate about the peak of the memory cycle could arise.

The fact that demand is strong and the judgment that stock prices can continue to rise are not the same. As much as the magnitude of demand, the duration of demand, the pricing power, and the speed of investment by customers have become important.

Three structures were confirmed in this week's market.

First, AI demand remains strong.

The strength of Dell, Nvidia, Hewlett-Packard (HP), and software stocks shows that AI investment is expanding beyond hardware to applications and platforms.

Second, AI hardware leaders are no longer unconditionally safe.

The sharp declines in Broadcom and Micron demonstrated that even strong earnings may not be sufficient to defend stock prices when high expectations are coupled with asset lifespan controversies.

Third, the gap between Big Tech and the lower tiers of the value chain is likely to widen.

Companies with massive cash flows and equity can absorb the burden of AI investment, while companies reliant on debt and collateral may face pressure on their financing costs as chip generational shifts accelerate.

The conclusion of this week's stock trends is as follows:

“The AI rally is not over, but the market where all stocks rise together under the name of AI is weakening.”

Next week's observation points are threefold:

First, we need to see if the Philadelphia Semiconductor Index and major Korean semiconductor stocks recover from their declines after the Broadcom shock.

Second, we need to see if the debate about the peak of memory demand actually translates into pressure on the supply and demand for Samsung Electronics and SK Hynix.

Third, we need to check if the strength in AI PCs and software expands into a new leading trend that fills the void left by semiconductor adjustments.

※ This article is a market commentary to aid in understanding the article and not investment advice. Actual market and stock prices may vary due to fluctuations in exchange rates, interest rates, oil prices, earnings, policies, and supply and demand variables.