In Johann Wolfgang von Goethe’s poem “The Sorcerer’s Apprentice,” a young assistant, tired of hard work, casts a spell to make a broom fetch water instead of him. When the task is complete and he does not know how to stop the magic, instead of waiting and looking for the correct spell, he cuts the broom in two. Each half becomes independent, and the problem doubles. The more he tries to control the situation by adding more, the more uncontrollable it becomes.
Today, we are witnessing something similar in the capital markets. Major technology companies and chip manufacturers continue to pour “fuel on the fire” of artificial intelligence. Instead of slowing down and waiting for clearer signals of return on investment, they are investing more aggressively. SK Hynix has just raised $26.5 billion on Nasdaq to build new factories. Its competitors are doing the same. The major cloud companies are increasing their capital expenditures by tens of percent every year. Every new commitment turns into another broom fetching water - or in this case, another billion dollars of capacity.
On Friday, July 10, 2026, SK Hynix raised $26.5 billion through American Depositary Receipts on Nasdaq - the largest-ever initial listing of a foreign company in Wall Street history. The offering was oversubscribed more than seven times, and the shares closed up approximately 13%. At a time when geopolitical risks in the Middle East were once again in focus, the market chose to pay a premium for exposure to the most critical link in the artificial intelligence supply chain - HBM chips (High-Bandwidth Memory).
Table 1. Key Data from SK Hynix’s Nasdaq Debut
| Metric | Value | Significance |
|---|---|---|
| Amount Raised | $26.5 billion | Second-largest equity offering after SpaceX |
| Oversubscription | More than 7x | Extremely strong institutional demand |
| First-Day Gain | ~ +13% | Strong vote of confidence from the market |
| HBM Market Share (2026) | 50-58% | Absolute leader in the critical AI segment |
Valuations of SK Hynix and Memory Makers: Still “Cheap” or Already Dangerous?
Paradoxically, despite the explosive growth in profits, memory stocks are still trading at relatively low multiples relative to growth:
Table 2. Forward P/E Multiples Comparison (July 2026)
| Company | Forward P/E | Comment |
|---|---|---|
| SK Hynix | 6.4 - 7.1x | HBM leader with re-rating potential |
| Samsung Electronics | ~6.4x | Large scale but lower HBM exposure |
| Micron Technology | ~9.2x | Highest multiple among the three |
AI Capital Expenditure Risk: When Everyone Builds at the Same Time
The biggest medium-term risk is not a lack of demand today, but oversupply tomorrow.
SK Hynix, Samsung, and Micron are all investing aggressively and simultaneously. SK Hynix and Samsung alone plan to spend hundreds of billions of dollars on new production capacity over the next decade. Goldman Sachs estimates total AI-related capital expenditures by the major cloud companies for 2025-2027 at approximately $1.15 trillion - more than double the previous three-year period.
Free Cash Flow Dynamics: Who Pays the Bill and Who Collects It
Behind the aggressive capital spending lies an even more important structural shift - a complete reversal of the historical free cash flow dynamics between the major cloud companies and semiconductor manufacturers.
According to Goldman Sachs and PIMCO estimates, AI-related capital expenditures now absorb around 94% of the operating cash flow of the major cloud companies over the next two years - up from about 40% in 2023.
At the same time, free cash flow at semiconductor companies is rising sharply. The money that hyperscalers spend on building data centers and purchasing chips is directly converted into revenue and profits for Nvidia, SK Hynix, Samsung, and Micron.
Comparison with Previous Cycles: The AI Investment Cycle (2025-2027) vs. the Dot-Com Boom (1999-2000)
Many analysts draw parallels with the late 1990s and the internet boom. There are similarities, but also important differences:
Table 3. Comparison Between the AI Investment Cycle (2025-2027) and the Dot-Com Boom (1999-2000)
| Aspect | Dot-Com Boom (1999-2000) | AI Cycle (2025-2027) |
|---|---|---|
| Capital Expenditures | Very High | Extremely High |
| Company Profits | Weak or Non-existent | Strong and Growing |
| Technological Maturity | Low | Higher |
| Valuations (Forward P/E) | Extremely High | More Moderate |
What Does SK Hynix’s Debut Mean for Investors in 2026-2028?
SK Hynix’s debut is a bullish signal for the next 12-18 months. Demand is real, margins are high, and the market has not yet paid the full premium for leadership in the HBM chip segment.
At the same time, it does not eliminate the structural risk for the 2027-2029 period. If the major cloud companies begin to optimize their capital expenditures - due to questions about return on investment, pressure on free cash flow, or macroeconomic slowdown - the memory sector could enter a classic downturn precisely when new production capacity comes online.
For investors, this means:
- Short term: Continued strength in companies tied to AI infrastructure (including memory leaders).
- Medium term: Greater selectivity. Companies with better capital expenditure discipline, long-term contracts with guaranteed payments, and diversification will perform better.
- Valuations in the memory segment still offer relative attractiveness compared to growth - but this also reflects skepticism about the duration of the boom.
Conclusion
The Sorcerer’s Apprentice is ultimately saved by his master, who casts the correct spell and restores order. In the capital markets, however, there is no such omnipresent sorcerer. There is no central force that can say “stop” to the arms race in AI capital expenditures when it starts to spiral out of control.
SK Hynix and other players in the memory segment currently enjoy strong demand and high margins. They are adding more fuel to the fire because the market pays them to do so. But the history of semiconductor cycles, as well as Goethe’s lesson, reminds us that every new “spell” - every new factory, every new multi-billion-dollar commitment - makes future control of the fire more difficult.
The question is no longer whether the fire of artificial intelligence will continue to burn. It is already burning strongly. The question is who will have to cast the stopping spell when the flames become too high - and whether anyone will still know how to do it by then.
Frequently Asked Questions (FAQ)
1. What exactly is HBM memory and why is it so critical for artificial intelligence?
HBM (High-Bandwidth Memory) is high-speed, multi-layered memory placed directly next to graphics processors. It enables significantly faster data transfer compared to traditional memory. In the context of artificial intelligence, HBM is critical because AI models process enormous volumes of data simultaneously. Without sufficient HBM capacity, even the most powerful chips cannot operate at full performance.
2. Why was SK Hynix able to raise $26.5 billion on Nasdaq despite geopolitical risks?
The market clearly showed its priorities. Despite tensions between the United States and Iran, investors valued the long-term potential of artificial intelligence more highly than short-term geopolitical risks. SK Hynix is the absolute leader in the HBM segment with a 50-58% market share, and demand from major technology companies remains very strong.
3. What is the risk of HBM chip oversupply in 2027-2028?
The risk is real. SK Hynix, Samsung, and Micron are all investing aggressively and simultaneously. If demand from the major cloud companies slows in 2027-2028, the new production capacity could lead to oversupply. This is a classic scenario for the memory industry - when all players increase capacity at the same time, prices and margins often contract sharply.
4. How does the reversal of free cash flow affect future AI capital expenditures?
This is one of the most important structural shifts currently underway. The major cloud companies are now directing a very large portion of their operating cash flow toward AI capital expenditures. As a result, their free cash flow is declining, and they are increasingly resorting to debt financing. If this pressure intensifies, companies may become more conservative with new investments, which will directly impact the revenues of chip manufacturers.
5. Could this AI investment cycle turn into another dot-com bubble?
There are similarities, but also significant differences. Similar to the late 1990s, we see massive capital expenditures in infrastructure and strong performance by equipment and chip suppliers. At the same time, today’s companies are already generating real revenue from artificial intelligence, have stronger balance sheets, and more mature technology. Valuations are also not as extreme as in 1999-2000. However, the risk of a correction still exists, especially if the major cloud companies slow their capital spending or if oversupply emerges in the HBM segment.
Important notice
This article is for information and education only. It is not financial advice, an investment recommendation or professional advice of any kind. The views expressed are the author’s personal opinions and are not recommendations to buy, sell or hold securities, assets or investments. Investing involves substantial risk, including the possible loss of all capital. Past performance is not a guarantee of future results. Each reader should conduct independent research, consult a qualified financial adviser and weigh financial circumstances, risk tolerance and investment objectives before taking any decision. The author and borsite.net accept no liability for any loss arising from the use of the information in this article.