What is AI bubble?
An AI bubble is a situation in which expectations for AI-related companies are so high that stock prices and corporate values rise faster than growth in sales and profits.
For example, it is used in the following situations:
| Status | What's happening |
|---|---|
| AI-related stocks soar | Expectations for future growth are the first to ride on stock prices |
| Funds are attracted even to immature companies | Even if profits are low, they are evaluated based on their theme |
| PER and PSR become high | Stock price appears expensive relative to current profits and sales |
| Even companies with little connection to AI are being bought | Funds come in just by saying "AI-related" |
Bubble is a strong word. So, be a little careful when using it.
This does not mean that AI itself is empty. In fact, AI as a technology is starting to enter real business. The key to determining whether there is a bubble is not the value of the technology, but the price investors are paying.
Is the technology real?
↓
Will the company make money?
↓
Are stock prices too high?Although these three questions seem similar, they are different questions.
Why is it called an AI bubble?
There is a large influx of capital into AI.
According to Stanford HAI's 2026 AI Index, global corporate AI investment in 2025 is estimated to be $581.7 billion, an increase of 130% from the previous year. Private investment also increased significantly from 2024 to $344.7 billion.
With so much money gathering, AI-related stocks are likely to stand out in the stock market.
Representative themes are as follows.
| Field | Relationship with the AI boom |
|---|---|
| AI semiconductors | Supplying GPUs, dedicated semiconductors, and memory necessary for learning and inference |
| Data center | Places an AI server and supports large-scale calculation processing |
| Cloud | Becomes the foundation for companies to use AI |
| Software | Provides AI functions, business automation, and development support for companies |
| Power/Cooling | Related to data center power demand and heat countermeasures |
Up to this point, we are talking about actual demand.
However, in the stock market, themes with real demand tend to be bought in advance. Growth in several years' time is factored into the current stock price at once, so if there is even a slight slowdown in growth, the stock price may be adjusted significantly.
The term "AI bubble" refers to a situation in which "genuine demand" and "stock prices that are too proactive" are mixed.
Difference between bubble and growth theme
Growth themes and bubbles look similar from the outside.
In both cases, stock prices tend to rise, news coverage increases, and they attract the attention of investors.
The difference lies in whether profits and cash flow are keeping up.
| Perspective | Growth themes | Conditions that can easily become a bubble |
|---|---|---|
| Sales | Actual growth | Expectations are ahead |
| Profit | Profit margin is also improving | Sales are increasing but there is no profit left |
| Stock price | Rise in line with growth | Rise faster than profit growth |
| Fundraising | Used for growth investment | Funds are collected just by the theme name |
| Investor psychology | Buying based on numbers | Buying due to fear of missing out |
For example, let's say that corporate profits grow by 20%, but stock prices rise by 100%.
If the future of the company really leads to 5x or 10x profit growth, it may be possible to explain. However, if the basis for this is weak, it would be easy to see it as a promising move.
The scary thing about bubbles is that they can happen even to good companies. Even if the company is growing, if the stock price rises too high, investment returns will be poor.
Difference from IT bubble
The AI boom is often compared to the IT bubble around 2000.
There are some similarities.
- Expectations were raised that new technology would change society.
- Funds were concentrated in related companies
- Companies with small profits were also highly evaluated
- Investors' fears of missing out were likely to increase
However, there are some differences.
| Item | IT bubble | AI boom |
|---|---|---|
| Stage of technology dissemination | Early stage of Internet dissemination | Generative AI and cloud AI already in use |
| Monetization | Many companies had no profits | Some large companies generated huge sales and profits |
| Infrastructure | Communication networks and EC infrastructure are expanding | Cloud, semiconductors, and data centers already exist |
| Investment burden | Focused on communications and internet companies | Expands to semiconductors, electric power, and data centers |
In other words, the AI boom is not simply ``dangerous because it lacks substance.''
In fact, it is precisely because there is real demand that the market tends to become bullish. This is the difficult part. Even if the technology is genuine, stock prices may not always be correct.
Fields likely to benefit from AI-related topics
There are a wide range of fields that are attracting attention due to the AI boom.
semiconductors
Generative AI requires a large amount of computation.
Therefore, GPUs, AI accelerators, high-performance memory, semiconductors manufacturing equipment, advanced packages, etc. are attracting attention.
However, semiconductors are also highly cyclical. Profits tend to soar when demand is strong, but stock prices also fall when inventory adjustments and investment run their course.
data center
Learning and operating AI models requires facilities to house servers, electricity, cooling, and networks.
According to the IEA's "Energy and AI," data center investment in 2024 was approximately $500 billion worldwide, and data center power consumption was approximately 1.5% of global power consumption. It is also predicted that data center power consumption will double to approximately 945 TWh by 2030.
While this indicates high data center-related demand, it also means that power, grid, land, cooling, and zoning regulations can become bottlenecks.
cloud service
When companies use AI, they do not necessarily have to have a huge computing environment in-house.
In many cases, AI functions are used on cloud services. For cloud companies, this could lead to an increase in usage volume and contract unit prices.
However, capital investment for AI is also heavy on the cloud side. Even if sales increase, depreciation and electricity costs will increase first, making it necessary to check profit margins.
software
In corporate software, AI is increasingly used in document creation, inquiry response, code generation, sales support, accounting processing, security monitoring, etc.
What investors want to see is not just the addition of AI functionality.
Added AI function
↓
Usage rate increases
↓
Improve unit price and churn rate
↓
remain profitableCan you confirm this flow?
Even if the number of users increases, if the unit price does not increase and only the calculation costs increase, the market will gradually become calmer.
Main risks of AI bubble
Expected upfront risk
It is easy to factor in future growth in AI-related stocks.
Therefore, even if the financial results are not bad, a company may be sold simply because it is seen as ``not as good as expected.''
The stronger the theme stocks are, the more the problem is high expectations rather than good news.
Valuation risk
AI-related companies can be highly valued when looking at PER, PSR, EV/EBITDA, free cash flow yield, etc.
A high rating itself is not a bad thing. High-growth companies tend to command a certain premium.
However, justifying a high stock price requires not only sales growth, but also profit margins, cash generation ability, and sustained competitive advantage.
Capital investment burden
Although AI may seem like a light business, it is actually quite capital-intensive.
AI semiconductors, servers, data centers, power, cooling equipment, research and development, human resources recruitment. There is a lot of money to go out first.
It would be good if the investment would lead to future profits, but if the demand forecast is incorrect, the burden of excess equipment and depreciation becomes heavy.
Risk of increased competition
The AI market is populated by big tech companies, semiconductors companies, cloud companies, startups, and open source developers.
Intensified competition tends to lead to lower prices, higher customer acquisition costs, and commoditization of functions.
It is also important to consider the situation where simply being able to use AI is no longer sufficient to differentiate a company.
Regulatory and governance risks
AI has challenges such as privacy, copyright, discrimination, security, misinformation, and accountability.
NIST's AI Risk Management Framework has been published as a framework for incorporating trustworthiness into the design, development, use, and evaluation of AI. A profile for generative AI will be released in 2024, and the use of AI requires not only technology development but also risk management costs.
As regulations and internal controls become stricter, some companies' growth speed and profit margins will be affected.
Indicators that investors want to check
When looking at AI-related companies, you want to check not only the theme name but also the numbers.
| Indicators | Points to look at |
|---|---|
| Sales growth rate | Is AI demand actually reflected in sales |
| AI-related sales ratio | What percentage of total sales does AI account for |
| Gross profit margin | Can AI services and semiconductors maintain high profit margins |
| Operating profit margin | Does profit remain even after R&D expenses, personnel costs, and depreciation |
| Free cash flow | Does it generate cash in addition to accounting profits? |
| Capital investment | Growth investment or likely to be excessive investment |
| PER/PSR | How much expectations affect stock prices |
| Customer concentration | Are you too dependent on a few large customers? |
| Order backlog/contract period | Is there continuity of demand |
Especially for beginners, it is easy to feel reassured by looking only at sales growth.
However, in AI-related fields, there are many situations where profits are more important than sales, and cash is more important than profits. For companies with large capital investments, checking cash flow is essential.
Points that beginners tend to misunderstand
Anything related to AI will definitely go up.
AI is a big growth theme, but AI-related stocks will not necessarily go up.
Even if a company is good, if you buy it at too high a price, the return will be low. Stocks with strong themes may be sold in bulk to reduce the risk of the overall market.
The whole AI boom is a bubble.
This too is extreme.
Corporate use of AI, demand for semiconductors, cloud investment, and data center construction are actually progressing. If we dismiss AI as just a buzzword, we will overlook changes in the industrial structure.
What we need to look at is not just whether the AI is real or not. The question is which companies are in a position to make profits.
You only need to buy AI companies
Theme-focused investment is great if it's successful, but it also hurts greatly if it's not.
Even if you own AI-related stocks, semiconductors stocks, or theme-based ETFs, you need to consider the balance with global stocks, US stock indexes, cash, bonds, etc.
Even when holding NISA for a long time, there are still risks of loss of principal, exchange rate fluctuations, large fluctuations in valuation, and concentration risks in specific themes.
How to view the AI boom from a long-term investment perspective
In the long term, the AI boom has two faces.
short term
Anticipation lead, sudden rise, adjustment, theme cycle
long term
Productivity improvement, cloud computing, semiconductors demand, business automationLooking back at history, technologies such as railways, automobiles, electricity, the Internet, and smartphones have taken root in society, with some overheating and some disappointments along the way.
Similarly, AI may remain as a technology, but not everyone will win as an investment.
If you look at it from a long-term investment perspective, it will be easier to remain calm if you ask the following questions.
- Is the company increasing sales with AI?
- Are you improving not only sales but also profit margin?
- Is it likely that you will be able to recover your capital investment and R&D costs?
- Will pricing power remain even if competition increases?
- To what extent do stock prices incorporate future growth?
There's a difference between believing in the future of AI and justifying a high stock price right now.
If you can think about this separately, you will be less likely to be swayed by the term AI bubble.
summary
An AI bubble refers to a situation in which expectations for AI become so high that the stock prices and corporate values of related companies rise above their actual capabilities.
The points to keep in mind are as follows.
- There is real demand for AI, and it cannot be said to be just a speculative theme.
- Whether there is a bubble or not is determined not by the value of the technology but by incorporating it into the stock price.
- Benefits extend to AI semiconductors, data centers, cloud, software, etc.
- Capital investment, competition, regulation and power constraints pose major risks
- When investing, check not only sales growth but also profit margin, cash flow, and valuation.
AI is a technology that has the potential to change the economy and business activities.
Still, good technology doesn't necessarily mean it's a good investment. What investors want to see is not only the story of the future, but also how much is reflected in the current financial results, and whether the stock price is too high relative to those numbers.
reference
- Stanford HAI, “Inside the AI Index: 12 Takeaways from the 2026 Report,” reviewed June 18, 2026. https://hai.stanford.edu/news/inside-the-ai-index-12-takeaways-from-the-2026-report
- Stanford HAI “AI Index”, confirmed June 18, 2026. https://hai.stanford.edu/ai-index
- IEA "Energy and AI - Executive summary", confirmed June 18, 2026. https://www.iea.org/reports/energy-and-ai/executive-summary
- NIST “AI Risk Management Framework”, confirmed June 18, 2026. https://www.nist.gov/itl/ai-risk-management-framework
- NIST AI Resource Center "AI Risk Management Framework", confirmed June 18, 2026. https://airc.nist.gov/airmf-resources/airmf/