Summary of Key Points
Recently, there have been significant fluctuations in global semiconductor stocks, especially those related to memory chips: U.S. chip stocks have fallen by more than 11% in two weeks, with Samsung and SK Hynix experiencing substantial declines (Samsung's 6.8% drop even triggered a circuit breaker in South Korea). Morgan Stanley believes that investment trends are shifting from semiconductor stocks to "hyperscalers" such as Google and Amazon, which are the main investors in AI data centers. SK Hynix is moving forward with an IPO in the U.S. (aiming to raise $28 billion), but there is disagreement about the sustainability of the AI boom. In the short term, there are risks such as rising memory prices and delays in data center construction; however, the long-term trend of increasing spending on AI infrastructure remains unchanged.
1. Why Have Semiconductor Stocks Suddenly Cooled Down? Is There a Shift in Investment Directions?
Why have chip stocks declined recently? According to Morgan Stanley, it's mainly due to "profit-taking." The previous AI boom caused chip prices to rise significantly, and investors are selling their positions once they have made enough profits, leading to the decline. More importantly, market sentiment is changing: while before investors were focused on companies that sell chips (like Samsung and SK Hynix), now the focus may be shifting to those that buy chips for data center construction—i.e., hyperscalers like Google, Amazon, and Meta.
What are hyperscalers? Simply put, these are tech giants with large-scale cloud services and data centers; they are the largest buyers of AI infrastructure (for example, building an AI data center requires purchasing a large number of memory chips). Chip stocks rose because investors expected these companies to increase their purchases. Now that chip prices have corrected, funds are moving towards these "chip buyers" as they are likely to benefit from the long-term expansion of AI infrastructure.
2. Samsung's Performance Has Increased 19-Fold, but Why Does SK Hynix Want to Go Public Now?
Samsung's situation is somewhat counterintuitive: its profits in the second quarter increased by 19 times year-over-year (nearly $58.4 billion), yet its stock price fell by 6.8%. The reason is simple—positive news was already anticipated by the market, and since there were no new surprises when the results were released, investors took the opportunity to sell.
SK Hynix is pursuing an IPO in the U.S. (aiming to raise $28 billion, second only to SpaceX in scale). The rationale behind this move includes: ① expanding its investor base globally and enhancing international competitiveness; ② the market's long-term optimism about AI demand (although there are short-term fluctuations, investors believe that demand for SK Hynix's products will remain strong in the U.S. stock market); ③ the potential to be included in the Philadelphia Semiconductor Index after listing (which would provide it with "quality stock" status and attract more investment. HSBC has even raised its valuation of SK Hynix by 20%, indicating institutional confidence in its long-term prospects.
3. Is the AI Boom About to Cool Down? Short-Term Challenges, but Long-Term Potential Remains Strong
There are two perspectives on the sustainability of the AI boom:
Short-Term Risks:
- Rapid increases in memory chip prices may lead companies to reduce spending on AI infrastructure and consumer electronics (as higher chip costs mean fewer purchases).
- Delays in U.S. data center construction (due to labor shortages and power constraints) could temporarily decrease chip demand.
- Analysts suggest that the "supercycle" for memory chips has passed its initial phase, indicating that prices may stabilize after the recent surge.
Long-Term Optimism: Goldman Sachs predicts that the combined investment of Meta, Microsoft, Amazon, and Google in AI-related areas will reach $5.3 trillion by 2025-2030—this is an enormous figure! This shows that the long-term demand for AI infrastructure will not decline, although the pace may slow down in the short term.
4. Hyperscalers Become the New Focus: They Are the Real Drivers of AI Infrastructure
Why are these giants suddenly being favored? Because they are the ultimate buyers of AI infrastructure. For example, building an AI data center requires purchasing chips, servers, and software, but who actually pays for all that? It's companies like Google and Amazon.
Morgan Stanley believes that when semiconductor stocks experience fluctuations, hyperscalers will benefit because they control the pace of AI infrastructure development and can pass on cost changes (such as price increases) to their customers (e.g., by raising prices for their cloud services). Additionally, these companies have more stable businesses and are less affected by economic cycles compared to chip manufacturers.
Conclusion
In the short term, chip stocks may continue to fluctuate, but the long-term demand for AI infrastructure is clear. Investors are shifting their focus from companies that produce chips to those that utilize them in practical applications (hyperscalers). Ultimately, it's the giants that will profit significantly from the long-term development of AI. SK Hynix's IPO and Samsung's stock price fluctuations are just small parts of this larger trend.
(The entire text is explained in plain language, making it easy for non-financial professionals to understand.)