🚨 AI Stocks Surge 230% in 3 Years
Today Korean Economic News for Beginners | 2025.11.13
0️⃣ Shifting from 'Investment Expansion' to 'Profit Realization' Era
📌 Capital Investment Reaches Peak… Profitability Speed Determines Stock Prices, Power Shortage and Financing Emerge as New Variables
💬 Over the past three years, AI-related stock prices in major countries have surged over 230%, recording remarkable growth. However, from next year, 'profit realization speed' rather than investment scale is expected to determine stock prices. As global major tech companies' data center and semiconductor facility investments reach their peak, how much actual profit they generate is emerging as the key criterion for evaluating company value. While semiconductor and cloud infrastructure companies are already making profits, AI platform and application companies are still struggling with losses, and their fortunes are expected to diverge based on profitability speed. Meanwhile, power shortages are intensifying due to high power consumption of AI servers, and corporate bond issuance is surging to raise massive investment funds, making financial stability and energy efficiency new competitiveness indicators.
1️⃣ Easy to Understand
AI stocks have exploded over the past three years. But now the market is starting to focus more on 'how much money companies actually make' rather than 'how amazing the technology is.'
Let's look at the numbers first. Over three years from 2022 to 2025, AI-related stocks rose an average of over 230%. Simply put, if you invested 10 million won, it would have become over 33 million won. Major tech companies like Nvidia, Microsoft, and Google led this rise.
Why did they rise so much? The AI boom started when ChatGPT appeared in late 2022. Companies competed to invest in AI technology, and investors bought related stocks with the expectation that "AI will change the world." Indeed, many companies poured massive amounts of money into building AI data centers and buying the latest semiconductors.
For example, Microsoft invested over 50 trillion won in AI infrastructure last year alone. Google, Amazon, and Meta increased investments on a similar scale. These investments benefited everyone from Nvidia making semiconductors, to Dell providing servers, to construction companies building data centers.
But now the situation is changing. The market is starting to say, "We've invested enough, now it's time to make money." Experts predict that from 2026, 'profitability speed' rather than investment scale will determine stock prices.
What is profitability? Simply put, it means money invested in technology or services actually returning as sales and profits. If you spent 10 billion won developing an AI model, you need to sell it and earn 15 billion won for successful profitability.
Currently, the AI industry can be divided into three layers.
The first layer is infrastructure companies. Companies like Nvidia and TSMC that make semiconductors, and companies like AWS and Microsoft Azure that provide cloud services belong here. They're already making profits. Because semiconductors and cloud servers are essential for AI, demand has exploded, and they're making big profits.
For example, Nvidia's sales last quarter increased over 200% year-over-year. This is because their GPUs (graphics processing units), which are AI semiconductors, sold like hotcakes. Microsoft also made big profits as its Azure cloud business grew rapidly.
The second layer is AI platform companies. Companies developing large AI models like OpenAI (which made ChatGPT), Google's Gemini, and Meta's Llama. They're making massive investments now but profits are limited.
Take ChatGPT as an example. It offers a paid subscription service at $20 per month (about 28,000 won), but the server cost to answer one question is several cents. The more users there are, the more costs increase too. OpenAI is expected to lose $5 billion (about 7 trillion won) this year. They've made massive investments but profits haven't caught up yet.
The third layer is AI application companies. Startups creating specific services using AI. There are various fields like image generation AI, writing AI, and customer service chatbots. Most of these are still losing money. Technology development costs a lot, but it takes time to acquire customers and increase paid subscribers.
So why has 'profitability' become important now? There are two main reasons.
First, investment has reached its peak. Major tech companies have poured enormous amounts of money over the past 2-3 years, but they can't keep increasing investment indefinitely. Shareholders have started pressuring them, asking "How long will you just spend money? Shouldn't you make profits now?" In fact, some major tech companies have announced they'll reduce or freeze next year's AI investment compared to this year.
Second, interest rates are still high. When investors invest money in companies, they can get a guaranteed 4-5% return by investing in safe government bonds. But investing in AI companies has uncertain returns and high risk. In this situation, investors want to invest only in "companies that can really make money." Stock prices no longer rise just on vague expectations for the future.
Two new problems are additionally emerging here.
The first is the power problem. AI servers consume tremendous amounts of electricity. The power consumed by one large data center equals that of a small city. Asking ChatGPT one question uses over 10 times more electricity than a regular Google search.
Power shortages are intensifying as power demand explodes worldwide. In some parts of the United States, new data center construction is being delayed due to power shortages. Companies are considering restarting nuclear power plants or building their own power generation facilities. But all of these involve additional costs. If power costs rise, AI service profitability worsens further.
The second is the financing problem. Massive investment requires massive money. Major tech companies have lots of cash on hand, but it's not enough, so they're issuing corporate bonds to borrow money. The problem is that borrowing money when interest rates are high means a heavy interest burden.
For example, suppose Company A issued 10 trillion won in corporate bonds for AI investment. If the interest rate is 5%, they must pay 500 billion won in interest every year. If the AI business doesn't make profits, even this interest becomes difficult to bear. Startups with low credit ratings have to pay even higher interest rates, so their burden is greater.
In the end, these questions become important going forward: "Can this company actually make profits with the money invested?" "Can they make profits while bearing power costs?" "Do they have a stable enough revenue structure to repay debts?"
Infrastructure companies are already answering these questions. Nvidia shows high profitability, and cloud companies are creating stable cash flows. But AI platform and application companies haven't proven themselves yet.
In the end, AI stock investment is now shifting its center of gravity from 'technology competition' to 'profit competition'. Companies that actually make money with their technology, rather than companies with cool technology, will be the winners.
2️⃣ Economic Terms
📕 Capital Expenditure (CapEx)
Capital expenditure is money a company spends to acquire or expand long-term assets.
- It means money invested in assets to be used long-term, like factories, facilities, data centers, and semiconductor production lines.
- In the AI industry, typical capital expenditures include data center construction, GPU purchases, and server expansion.
- High CapEx means active investment for the future, but it can also burden short-term profitability.
📕 Monetization
Monetization is the process of converting money invested in technology, products, or services into actual sales and profits.
- Typical monetization methods include making free services paid, generating advertising revenue, or introducing subscription models.
- In the AI industry, the key to monetization is selling developed AI models as APIs or converting them to paid subscription services.
- The faster the monetization speed, the more positively investors evaluate the company.
📕 Return on Investment (ROI)
Return on investment is an indicator showing how much profit was generated relative to the amount invested.
- ROI = (Profit from investment / Investment amount) × 100.
- For example, if you invested 10 billion won and earned 12 billion won in profit, the ROI is 20%.
- Companies with high ROI in the AI industry are using capital efficiently and have high investment attractiveness.
📕 Corporate Bonds
Corporate bonds are bonds issued by companies to raise funds, promising to pay principal and interest to investors.
- It's a way for companies to raise funds directly from investors by issuing bonds instead of bank loans.
- Companies with higher credit ratings can issue corporate bonds at lower interest rates.
- AI companies are increasing corporate bond issuance to raise massive investment funds, but interest burden and repayment risk are emerging as new challenges.
3️⃣ Principles and Economic Outlook
✅ Phased Transition of Growth Industries
All innovative industries transition from the initial investment stage to the profitability stage, and the AI industry is now at that turning point.
First, we need to understand the S-curve theory of technological innovation. When new technology emerges, investment and interest initially increase explosively. The internet did this, smartphones did this, and now AI is doing this. In the early stage, stock prices rise just on expectations about technology's potential. Investors pour money in believing "this technology will change the world." But after a certain point, the market becomes cold. Questions start arising like "Does it really make money?" "When will it make profits?" The AI industry is at exactly that point now. From 2026, profit realization speed rather than investment scale will be the key factor determining stock prices.
Second, we must remember the lessons of the dot-com bubble. In early 2000, internet companies grew explosively. Stock prices of companies like Amazon and eBay soared sky-high. But numerous dot-com companies with unclear profit models also surged together, and when the dot-com bubble burst in 2000, most disappeared. Surviving companies were only those with models that could actually make profits, like Amazon. The AI industry now shows similar patterns. The era when you could get investment just by attaching the name 'AI' is ending. The gap between companies that actually make money and those that don't will widen increasingly.
Third, investment efficiency becomes the new competitiveness. If in the past "how much you invest" was important, now "how efficiently you invest" becomes important. If Company A creates 12 billion won in value and Company B creates only 8 billion won in value with the same 10 billion won investment, investors will naturally choose Company A. AI infrastructure companies receive high evaluations because their profit relative to investment is clear. Meanwhile, AI startups with unclear profit models are starting to face difficulties in attracting investment.
As the industry enters maturity, the transition from 'the age of technology' to 'the age of profit' is accelerating.
✅ Widening Gap in Profitability Speed
Even within the AI industry, the gap between companies that have succeeded in profitability and those that haven't is rapidly widening.
First, the infrastructure layer has already succeeded in profit generation. Nvidia's GPUs are so popular that supply can't keep up with demand. Because they're essential components for AI, they have high pricing power and good margins. Cloud companies are the same. Microsoft Azure, Amazon AWS, and Google Cloud all saw sales surge thanks to AI demand and maintain high profitability. They've already recovered their investment costs and are creating stable cash flows. Therefore, their stock prices are relatively stable and show steady growth.
Second, the platform layer is struggling with profitability. OpenAI has secured tremendous users with ChatGPT but is still losing money. While charging $20 per month for paid subscriptions, server operating costs are enormous. Moreover, they're under price reduction pressure as competition intensifies. Google's Gemini and Meta's Llama face similar situations. They must bear enormous development and operating costs while simultaneously making profits, which isn't easy. These companies are trying to increase profits in the B2B (corporate customer) market, but it's not enough yet.
Third, the application layer is in the most difficult situation. Numerous startups in image generation AI, writing AI, voice synthesis AI, etc. have emerged, but most are failing to establish profit models. When services are provided for free, users flock but there's no money. When they monetize, users drop sharply. Even when trying to introduce advertising models, revenue is minimal because user numbers aren't sufficient. Eventually, many AI startups are suffering from funding difficulties, and some are closing or being acquired. Surviving companies are only those that have secured clear target customers and markets with willingness to pay.
Going forward, polarization within the AI industry will intensify, and only a few companies that succeed in profitability will be recognized for high value.
✅ Emergence of New Constraints
Two new constraints—power supply and financing—may limit the growth of the AI industry.
First, power shortage is becoming a real problem. The power consumed by one AI data center is beyond imagination. A large data center consumes hundreds of megawatts of power annually, equivalent to the power used by an entire small or medium-sized city. The problem is that power supply worldwide can't keep up with demand. In some parts of the United States, power grids have reached saturation, delaying or denying permits for new data center construction. Companies are pursuing nuclear power plant restart, renewable energy investment, and building their own power generation facilities, but all of these require additional cost and time. If power costs rise, AI service operating costs also rise, ultimately worsening profitability further.
Second, financing difficulties are growing. AI investment is enormous in scale. Building a single data center alone costs several trillion won. Major tech companies have lots of cash on hand, but it's not enough, so they're issuing corporate bonds. The problem is that borrowing money when interest rates are still high means a heavy interest burden. Especially small and medium-sized AI companies with low credit ratings must endure even higher interest rates, and some find financing itself difficult. Venture capital AI investment is also on a declining trend. In the past, you could get investment with just a good idea, but now it's difficult to attract investment without a clear profit model and financial plan.
Third, these two constraints will restructure the industry. Large companies that can secure power and funding will become stronger, while small and medium companies that can't are likely to be weeded out. Eventually, the AI industry may also become large-scale and oligopolistic. Just as the cloud market is dominated by three companies—Amazon, Microsoft, and Google—the AI market may also be dominated by a few large companies. This also raises concerns that it could harm innovation diversity.
For sustainable growth of the AI industry, expanding power infrastructure and establishing efficient financing mechanisms are essential.
4️⃣ In Conclusion
The 230% surge in AI stocks over three years is the result of expectations for technological revolution. But now the market is turning its eyes from expectations to reality, from investment to profit.
From 2026, 'how quickly profits are made' rather than 'how much is invested' will be the key criterion determining company value. Infrastructure companies have already passed this test. Nvidia, TSMC, and cloud companies are making stable profits, and stock prices are moving accordingly.
But AI platform and application companies haven't proven themselves yet. Despite massive investment, they're still losing money or have low profitability. Going forward, these companies will have diverging fortunes based on profitability speed. Companies that quickly establish profit models will survive and grow more, while those that don't will be weeded out.
With power shortages and financing as new constraints added, the future of the AI industry is becoming even more uncertain. Only companies that secure power and use it efficiently, and can raise stable funds with sound financial structures, will be able to survive long-term.
From an investor's perspective, how should you respond? The most important thing is not to invest just looking at the name 'AI', but to analyze actual revenue structures. You need to carefully examine whether this company can really make money, whether the return on investment is reasonable, and whether the financial structure is sound.
Infrastructure companies are relatively safe. They're already making profits, and stable growth is expected as long as AI demand continues. But if you invest in platform and application companies, you must take on high risk. You might make big profits, but you might also lose all your investment.
If you're a young professional or financial beginner, you should be even more careful. AI stock volatility is very high. It's common for prices to rise or fall over 10% in a single day. Never invest your living expenses or emergency funds. It's advisable to invest only a portion of your spare money, and even then, from a long-term investment perspective.
In the end, the AI revolution will continue. But companies and investors who benefit will be selected. 'The age of technology' is ending and 'the age of profit' is beginning. Understanding and responding properly to this change is the key to successful investment.
View Table of Contents