[Summary]
The AI market in 2026 will have a similar frenzy to the IT bubble of 2000, but its essence will be different. Currently, the focus is on companies that actually generate huge profits, such as Nvidia and Microsoft, and demand for AI is also accompanied by real demand in areas such as data centers, semiconductors, and electricity. On the other hand, the market has begun to overprice future growth, and caution is needed to prevent overheating in capital investment and declining profit margins. In particular, in the future, companies will no longer be evaluated solely for being ``AI-related,'' and only companies that can realize profits, cash flow, and return on investment will be the winners. Even if AI is real, stock prices are not always correct - this is the core of the current AI market.
Looking at the stock market in 2026, I am reminded of the IT bubble around 2000.
AI is too strong a word.
AI is also used at financial results briefings. AI is also used in investor materials. Semiconductors are also AI. Electricity is also AI. The data center is also AI. Air conditioning is also AI.
There are even situations where stock prices are bought simply by saying that it is ``related to AI.''
In fact, in today's market, AI is not just a technology theme, but a ``huge capital circulation''.
Nvidia, Microsoft, Amazon, Meta, Alphabet, TSMC, Broadcom.
Companies around here are no longer only operating in the software market.
GPU. HBM. Optical communication. Liquid cooling. transformer. Electric wire. Electricity. data center. Semiconductor manufacturing equipment.
AI is now transforming into an "infrastructure investment theme."
That's why many investors are worried.
“Isn’t this the same as the IT bubble after all?”
This question is quite important.
However, in conclusion, I think that they are not completely the same.
Rather, the current AI market is
A market where “genuine technological revolution accompanied by real demand” and “bubble formation due to excessive expectations” are progressing at the same time
It is.
If you don't understand this sloppily, it's quite dangerous.
That's because AI is real and being able to win no matter what price you buy AI stocks are two completely different things.
First of all, I would like to clarify that the IT bubble of 2000 was not caused by the Internet being a lie.
In fact, it's the opposite.
The Internet has completely changed the world since then.
search. E.C. SNS. Cloud. Smartphone. Video distribution. advertisement.
Today's economy cannot exist without the Internet.
In other words, the vision of the future that investors saw at the time was quite correct.
The problem was stock prices.
Although many companies in the market at that time were not making profits,
- Number of PVs
- Number of members
- Number of accesses
- “A market that will become huge in the future”
That alone justified the stock price.
As a result, NASDAQ collapsed.
However, the important thing here is that
"Technological revolution" and "investment return" were different things
That's the point.
Amazon has become a long-term winner. But investors who bought at the 2000 high suffered for a long time.
Cisco has also become a company that supports the world. However, after the bubble burst, stock prices were unable to rise above their highs for a long period of time.
This is the essence.
Even good companies Even with good technology, If you buy at a high price, your investment return will be poor.
This is exactly the same in the AI market.
However, it is wrong to dismiss the current AI market as ``just a bubble.''
The biggest difference from 2000 is the quality of the leading companies.
Many of the dot com companies at the time were
- Deficit
- Small scale *Funding dependent
- No profit
It was.
However, the companies at the center of the current AI market are different.
Nvidia. Microsoft. Meta. Amazon. Alphabet. TSMC.
They are already making huge profits.
Moreover,
*Cash flow
- Customer base *Technical ability
- Network effect
- Data amount
I have up to
In other words, this time,
Not “enthusiasm without profits” but “excessive expectations for profitable companies”
What is it?
This is a pretty big difference.
In fact, current AI investments are not driven solely by expectations.
In reality, we don't have enough GPUs. Not enough HBM. Not enough electricity. Not enough data centers.
In other words, there is real demand.
This is a crucial difference from 2000.
What is important to understand the current AI market is:
“AI application competition”
It's not just about looking.
What is really important is the "infrastructure war" that is occurring behind the scenes.
Generative AI is incredibly computationally resource intensive.
ChatGPT, Gemini, and Claude all run on huge GPUs.
What happens then?
First, demand for GPUs will explode.
This will increase the demand for HBM.
As a result, demand for advanced packaging will increase.
This will necessitate the construction of a data center.
Then you will need electricity.
Then you will need a transformer.
Cooling equipment will then be required.
In other words, while AI is a software revolution, it is also a ``heavy, long, and large-scale capital investment cycle.''
Once you understand this, the way you view Japanese stocks will change considerably.
View the AI market price from three perspectives: “semiconductors, power, and profit structure”
When considering AI-related Japanese stocks, many investors tend to judge only by the word "AI-related."
However, in reality, even with the same AI theme, each company has completely different ways of making money.
If you invest without sorting things out, you may be able to get on board with the theme, but you won't be able to get the profits.
The following three perspectives are important here.
① Companies that will benefit from the semiconductor cycle
First and foremost is semiconductor manufacturing equipment and inspection equipment.
The representative is
- Tokyo Electron
- Advantest
- SCREEN Holdings
- Disco
This area is likely to benefit as long as the production of GPUs and HBM for AI continues to increase.
In particular, AI semiconductors are highly difficult to manufacture, and the demand for inspection processes and cutting-edge processes is increasing, so this is an area where Japanese companies tend to have a technological advantage.
However, there are some caveats.
This is also a typical "capital investment cycle stock."
In other words,
Increase in demand ↓ Increase in capital investment ↓ Rapid increase in orders ↓ Market expects permanent growth ↓ Reverse rotation due to oversupply
This trend is likely to occur.
Therefore, the most important thing is not the "AI theme", but
- backlog *Customer CapEx
- Maintain profit margin
- Stock circulation
It is.
② Companies that control the “power infrastructure” in the AI era
The next most important area is power and communication infrastructure.
This is surprisingly easy to overlook.
However, in reality, the main body of the AI market can be said to be "power consumption."
GPUs use electricity at abnormal levels. Moreover, it generates a large amount of heat.
Then, what you need is
- Electric wire
- Optical fiber
- Substation equipment
- Cooling
- Power distribution
It is.
What tends to attract attention in this area is
- Fujikura
- Furukawa Electric
- Sumitomo Electric Industries
- Hitachi, Ltd.
It is.
In particular, high-speed communication and large-capacity power supply are essential for AI data centers.
Therefore, in the future, not only "semiconductor companies" but also
“A company that supports the power grid in the AI era”
may become significantly more important.
③ Companies that survive in the long term are “companies that remain profitable”
And here is the most important thing.
In the AI market, there are times when stock prices go up solely due to sales growth.
But in the end, what the market sees is
“Will the company have any profits left?”
It is.
For example,
- Competition lowers prices
- Heavy capital investment burden
- Strong customer dependence
- Profit margin decreases
In this case, stock prices will suffer even if demand for AI grows.
On the contrary, the strong
- High share *Technical barriers
- High operating profit margin
- High FCF
- Strong pricing power
A company with
Seen from this perspective,
- Shin-Etsu Chemical *Keyence
- Daikin Industries
``Companies that are modest but have a strong profit structure'' like ``Companies with a strong profit structure'' are quite important.
In the end, the last thing left in the AI market is
"AI-related companies"
Not.
“A company that can maintain high profits even in the AI boom”
It is.
If you read this far,
“So, is it safe to buy AI stocks?”
Some people may think so.
But that's the danger.
Rather, the current market
“It’s easy to overheat because it’s real”
condition.
The market is already
“Will AI grow?”
rather than
"Isn't that growth already factored into the stock price?"
I'm starting to shift my focus to.
It is dangerous to misunderstand this.
Many people
“The demand for AI will suddenly disappear”
I worry about that.
But actually, it's not that scary.
The use of AI itself will likely continue to increase.
The problem is different.
What's really scary is
“Doubts about investment recovery”
It is.
For example, big tech companies are currently investing in AI at an alarming rate.
Buy a GPU. Build a data center. Secure electricity.
But what the market is looking at is what comes next.
“Can you really recover that investment?”
Here it is.
If,
- The unit price of AI services will not increase
- Inference cost is too high
- Competition lowers prices
- Profit margin deteriorates
If that happens, the market will suddenly change its evaluation.
In other words,
Even if demand for AI grows, It is normal for AI stocks to go down.
This is quite important.
Personally, I am very cautious about this in the current AI market.
Semiconductor and infrastructure investment are, after all, cyclical industries.
When demand is strong, everyone increases production.
Then,
- Factory construction *Capital investment *Increase inventory
happens all at once.
However, at some point supply constraints will loosen.
Then, it suddenly begins to rotate in reverse.
This has happened many times in the semiconductor industry.
There is a very similar atmosphere in the current AI market.
GPU shortage. HBM shortage. Lack of electricity.
This sense of scarcity is driving further investment.
But what's scary is
“I am most bullish when there is a shortage.”
That's the thing.
The market begins to price on the assumption of eternal scarcity.
However, in reality, supply capacity will definitely increase.
At that time,
- ASP decrease
- Decrease in profit margin
- Inventory adjustment
- PER reduction
begins.
Moreover, stock prices will fall before business performance worsens.
This is the fear of cyclical stocks.
Another rather important issue is the index issue.
Many people
“The S&P500 is decentralized and safe.”
That's what I think.
But in reality, it is heavily AI-biased.
The proportions of the Magnificent Seven have become too large.
In other words, even if you think you are buying the S&P500,
In reality,
*Nvidia *Microsoft *Amazon
- Meta *Alphabet
It is almost like having a large amount of.
This is similar to 2000.
Of course, the quality of the companies is completely different.
But,
"Structure in which the index depends on a minority of stocks"
itself is quite dangerous.
If AI investment expectations collapse, the entire index will be affected.
This is a pretty important point.
Personally, I think the AI market price from here on is
“Anything related to AI will go up.”
I think it will end in a phase.
Rather,
“A market where only companies that can make a profit remain”
Become.
This place is quite large.
What will become important in the future is
*Sales growth *Profit margin *Cash flow
- Return on investment
- Pricing power
It is.
Simply being “AI-related” is not enough.
For example,
A company whose sales are increasing but cash is disappearing due to capital investment.
Companies that have demand for AI, but whose profit margins are deteriorating due to competition.
Companies like this are quite dangerous.
On the contrary,
- High share
- Strong pricing power
- High FCF *Customer lock-in
Companies with this are strong.
In other words, from now on,
“Quality of profit”, not “AI dream”
is asked.
This is really important.
Personally, I think the most important thing when looking at the AI market is:
“Who ultimately benefits?”
It is.
I think the use of AI itself will increase.
However, investment returns are different.
Will GPU companies win? Will cloud companies win? Will AI model companies win? Will the power company win? Will infrastructure companies win?
It's not completely visible here yet.
It was the same with the internet.
Internet usage has exploded. But many of the early dot-com companies disappeared.
There is a strong possibility that the same thing will happen with AI.
So the important thing is
“Do you believe in AI?”
Not.
The important thing is that
“Who can recover the value of AI as profit?”
It is.
If you make a mistake here, you will lose out on investment even if you take advantage of the technological revolution.
The AI market in 2026 will definitely be a real technological revolution.
This is quite likely true.
I think AI will change society.
However, stock prices are different.
The market always anticipates the ``correct future'' too much.
That's why it's so dangerous.
The lessons of 2000 are:
“The internet was wrong.”
That's not the point.
The real lesson is
“Even in the right future, high price grabbing will always occur”
That's what it means.
The current AI market prices also seem to be quite close.
So from now on, what's important is
Rather than being labeled as AI-related,
*Profit margin *Cash flow
- Return on investment
- Pricing power *Cycle resistant
is to see.
AI may be real.
However, because it is genuine, investors need to remain calm.
Who will win in the next few years?
“People who believed in AI”
rather than
“A person who continued to monitor the profit structure even in the midst of AI craze”
I think so.