[Summary]

As of May 2026, the Japanese stock market has finished announcing its financial results for the fiscal year ending March, and there is a sense that AI, semiconductor, and defense-related stocks are running out of material in the short term.

However, what is important here is not just looking at the financial results for 2026.

What institutional investors are really looking at is

EPS expansion for fiscal years ending March 2027 and March 2028

It is.

AI, semiconductors, and defense are not just trending topics.

In the background,

  • AI data center investment
  • Strengthen semiconductor supply chain
  • Defense power development plan
  • National investment in cyber/space/electronic warfare
  • Edge AI and power saving

There is.

The Ministry of Economy, Trade and Industry has announced a policy to provide public support of more than 10 trillion yen over the seven years until fiscal 2030 in the AI ​​and semiconductor fields, and encourage public and private investment of more than 50 trillion yen over the next 10 years.

In the defense field, the national policy is to strengthen defense capabilities by 2027.

In other words, what we should be looking at heading into 2027 is:

“When will orders turn into sales and profits” rather than “theme popularity”?

It is.

In this paper, we will organize the national policy triple themes of AI, semiconductors, and defense as performance transformation scenarios for the fiscal years ending March 2027 and March 2028.

Introduction｜Eliminate the noise of May 2026 and anticipate the EPS expansion period of 2027

As of May 2026, the Japanese stock market has finished announcing its financial results for the fiscal year ending March, and there is a feeling that the market is running out of material in the near term.

AI-related stocks are also showing signs of correction compared to the sharp rise from 2024 to 2025.

Semiconductor manufacturing equipment stocks, defense-related stocks, and data center-related stocks are also susceptible to fluctuations in supply and demand immediately after the financial results period in the short term.

However, you should not look at the market only in terms of short-term price movements.

Institutional investors with market price discovery capabilities are already looking at more than just 2026 performance.

Their focus is

*Operating profit for the fiscal year ending March 2027

  • EPS for the fiscal year ending March 2028
  • Turn order backlog into sales
  • Improve profit margin
  • Probability of achieving medium-term management plan

is moving to.

Why AI, semiconductors, and defense now?

The reason is clear.

These three are not temporary theme stocks, but

“Megatrend supported simultaneously by national budget and global infrastructure investment”

That's why.

What investors tend to overlook is the time lag between receiving an order and taking profits.

Companies whose demand accumulated from 2024 to 2026 will turn into sales and profits in the fiscal years ending March 2027 and March 2028 may be reevaluated at the next market price.

Chapter 1 | The main battlefield for AI and semiconductors is from manufacturing equipment to power and cooling infrastructure

Generative AI and data center investment are already central themes in the semiconductor market.

However, the focus after 2026 has changed a bit.

The main characters from 2024 to 2025 are:

*GPU *HBM

  • Semiconductor manufacturing equipment
  • Tip exposure/film formation/etching
  • Front-end equipment investment

It was.

However, the main battlefield from the fiscal year ending March 2027 to the fiscal year ending March 2028 will be

“Infrastructure that physically keeps the AI server running”

is starting to move to

Specifically,

  • Power supply
  • Cooling *Liquid cooling
  • High efficiency air conditioning
  • Optical communication
  • Power semiconductor
  • Post-process packaging
  • Edge AI parts

It is.

The more computing power an AI server has, the more power it consumes and the heat it generates.

Starting with NVIDIA's Blackwell generation, the thermal design of liquid cooling systems and the entire data center has become an important issue.

In other words, the focus of the market is

“Companies that make semiconductors”

From,

“A company that stably operates AI”

It is expanding to

This change is not the end for the semiconductor market.

Rather,

“2nd round of AI market”

It is.

Paradigm shift to 2027

As we move towards 2027, the market focus will begin to shift from just manufacturing equipment to the peripheral infrastructure that allows AI to operate stably.

In the next generation AI server,

  • High performance GPU *HBM
  • Optical communication
  • Power control
  • Cooling
  • Highly reliable parts

are required together.

The important thing here is that the bottleneck changes.

The bottleneck in 2024 was GPU supply.

The bottleneck from 2025 to 2026 was HBM and advanced packaging.

And towards 2027,

“Power, cooling, mounting, and high reliability components”

becomes more important.

AI looks like a software theme.

However, as an investment theme, it is more about physical infrastructure.

Chapter 2 | Defense sector time lag analysis

If you view defense stocks as simply an extension of value stocks, you may miss changes that will occur after 2027.

The essence of the defense business is

“There is a time lag of 1 to 3 years from receiving an order to recording sales.”

That's it.

Even if government budgets increase, this does not mean that companies will benefit immediately.

The general flow is as follows.

2023〜2025year
defenseKey point
defenseKey point

↓

2025〜2026year
Key point
Key point

↓

2027〜2028year
Key point
salesKey point

In other words, the profits of the defense sector that are visible in the financial results as of 2026 may be figures that have not yet reached full scale.

Documents from the Ministry of Defense and the Self-Defense Forces also clearly state a policy to strengthen Japan's defense capabilities by fiscal 2027.

According to Ministry of Finance data, the defense-related budget for FY2020 is 8.7005 trillion yen.

This is not a temporary theme, but a multi-year theme backed by the national budget.

Defense is not just about tanks and fighter jets

Modern defense is not only about classic equipment.

The future focus will be on

  • Drone
  • Drone swarm control
  • Cyber defense
  • Space/satellite communications
  • Electronic warfare
  • Radar *Sensor
  • AI image analysis
  • Defense Communications Network

It is.

Defense and AI completely overlap here.

Autonomous control, image recognition, anomaly detection, communication encryption, and satellite data analysis are all extensions of AI, semiconductor, and communication technologies.

So when looking at defense stocks in 2027,

“A company that can turn defense budgets into profits from AI, communications, and electronic components”

you need to look for.

Chapter 3 | Favorite candidates for 2027, 3 categories x 2 stocks

From here, we will organize the candidates that are likely to bring about a change in performance in the fiscal years ending March 2027 and March 2028 into three categories.

Please note that this is not a short-term buy or sell recommendation.

What you need to look at is the rate of change in business performance toward 2027, backlog of orders, improvement in profit margins, and degree of connection with themes.

1. Cooling and power infrastructure

Sanyo Denki (6516)

Sanyo Denki is a company that develops cooling fans, servo systems, uninterruptible power supplies, etc.

Heat countermeasures and stable operation are important for AI servers, communication equipment, and industrial equipment.

Sanyo Denki's cooling fans are characterized by high air volume, high static pressure, low power consumption, and low vibration, making them compatible with applications that require long-term operation, such as servers and communication equipment.

The issues for 2027 are:

  • Increased heat generation of AI server
  • Cooling demand for high-density servers
  • Demand for renewal of data center peripheral equipment
  • Power saving needs

It is.

When looking at the company, rather than deciding that it is for AI servers,

“To what extent can we capture the increasing demand for highly reliable cooling components?”

It is realistic to check.

Daikin Industries (6367)

Daikin Industries is a company with a strong reputation for home air conditioners.

However, an important investment theme is air conditioning, chillers, and cooling solutions for data centers.

Data centers operate 24 hours a day, 365 days a year, so air conditioning equipment is not just peripheral equipment.

Poor cooling efficiency increases electricity costs and affects server operational stability.

Daikin is developing cooling solutions for data centers, and the expansion of AI data centers is an important tailwind for the company's BtoB domain.

The outlook for 2027 is

“To what extent can AI data center air conditioning infrastructure be monetized?”

It is.

If the valuation ratio of not only residential air conditioning but also industrial air conditioning and data center cooling increases, the way things are viewed in the stock market will change.

2. AI semiconductor infrastructure/post-processing

Tokyo Electron (8035)

Tokyo Electron is a company that represents Japan's semiconductor manufacturing equipment.

The strength of our front-end equipment is already reflected in the market.

However, what will be important for 2027 is the expansion of demand for equipment as AI semiconductors become more sophisticated.

In AI semiconductors,

*HBM *3D lamination

  • Wafer bonding *Advanced packaging
  • Highly integrated

becomes important.

In its medium-term management plan and IR materials, Tokyo Electron indicates its response to technological requirements for next-generation devices and high value-added products.

The issues in 2027 are:

“Can we capture the demand for high value-added equipment not only for front-end processes but also for the increasing complexity of AI semiconductors?”

It is.

In the short term, it will depend on semiconductor market conditions and orders.

However, in the medium term, the depth of AI semiconductor investment will determine the company's profit margin.

Murata Manufacturing (6981)

Murata Manufacturing is a global company that manufactures electronic components, mainly MLCCs.

The stagnation of the smartphone market until 2025 has weighed on electronic component stocks.

However, as we move towards 2027, our perspective will change.

AI servers, AI PCs, AI smartphones, and edge AI devices require highly reliable and high-performance electronic components.

Murata Manufacturing also has connections in the context of device-side AI, such as edge AI and AI cameras for factories.

The important thing is that

  • Power stabilization for AI servers
  • Parts for edge AI terminals
  • Recovery of high-performance smartphones
  • Computerization of automotive and industrial equipment

It is.

The scenario for 2027 is

“Re-evaluation of electronic components, which are increasing in volume and increasing added value at the same time”

It is.

3. Defense x AI/Advanced communications

Mitsubishi Electric (6503)

Mitsubishi Electric is a company that is strong in FA, electric power, air conditioning, and social infrastructure, as well as in the defense and space fields.

Defense electronics, satellite communications, radar, and space-related systems will be key themes heading into 2027.

In February 2026, the company announced that it had received an order from the Ministry of Defense to develop the next generation defense satellite communications.

This project symbolizes the trend toward the integration of defense, space, and communications.

The outlook for 2027 is

“When will defense and space orders turn into sales and profits?”

It is.

In addition, if the FA division recovers, it will lead to an improvement in the company's overall profit margin.

Not just defense,

  • FA recovery
  • Power infrastructure
  • Defense/Space
  • Semiconductor equipment investment

must be viewed together.

NEC (6701)

NEC has a strong impression of being an IT services company, but it is also a company deeply involved in defense, aerospace, communications, and cybersecurity.

The NEC Group provides ICT solutions for government agencies in the aerospace and defense fields.

We also have exposure to defense communications, sensors, networks and software through NEC Aerospace Systems and NEC Network Sensors.

What is important for 2027 is

  • Cyber defense
  • Defense Communications Network
  • Space/satellite data
  • AI analysis
  • Highly reliable system for government offices

It is.

NEC's strength is that it can handle communication, AI, security, and system integration all at once, rather than just hardware.

The scenario for 2027 is

“Expansion of high value-added projects where national defense policy and digital infrastructure overlap”

It is.

Chapter 4 | 5 checklists to identify winning stocks

To identify candidates for performance transformation in 2027, don't just follow the theme name.

The following five points need to be confirmed in quarterly financial results and IR.

Check 1｜Is the order backlog at a record high?

Order backlogs are important for defense, heavy electrical equipment, manufacturing equipment, and air conditioning equipment.

Backlog is a candidate for future sales.

Especially in the defense sector, there is a time lag between the budget increase and the recording of sales.

Therefore, to read the profit for 2027, we need to look at the order backlog as of 2026.

Check 2｜Is it directly connected to AI power, cooling, and post-processing?

There are many companies that call themselves AI-related.

But what will be important in 2027 are companies that solve the physical bottlenecks of AI.

Specifically,

  • Power
  • Cooling *Liquid cooling
  • Optical communication *HBM
  • Post-process
  • Highly reliable parts

It is.

You should look for companies that solve the constraints of AI infrastructure, not just AI software stocks.

Check 3｜Are you seeing an improvement in the operating profit margin?

Even if sales increase, if profit margins are poor, stock valuations will be difficult to increase.

Strong companies in 2027 will be

  • Cost upfront period ends
  • Ratio of high value-added products will increase
  • Mass production effect
  • Profit grows faster than sales

It's a company.

This is called operating leverage.

Investors need to see improvement in operating margins as well as sales growth.

Check 4｜Does it have both overseas sales ratio and national policy demand?

There is a limit to the scale of growth if Japan's national policy alone is used.

On the other hand, companies that can connect to overseas data centers, global semiconductor investment, and defense and space demand have more room for growth.

The ideal is

“Companies supported by domestic national policies and outperforming by overseas demand”

It is.

The three themes of AI, semiconductors, and defense cannot be completed only within Japan.

It is important to be connected to capital investment around the world.

Check 5 | Does the medium-term management plan clearly talk about 2027?

A medium-term management plan is important when looking at market prices in 2027.

In particular, you should check

  • Sales target for fiscal year 2027 *Operating profit margin target
  • ROE target
  • R&D investment *Capital investment
  • Priority areas related to defense, semiconductors, and AI

It is.

Companies whose management sees 2027 as a payback year will have higher resolution performance transformation scenarios.

Chapter 5 | Three possible risks

This scenario also has risks.

The stronger the theme, the greater the stock price adjustment if expectations are too far ahead.

Risk 1 | AI investment plateau

Big tech companies' data center investments may temporarily adjust.

AI investment is a long-term trend, but it doesn't grow straight every quarter.

If the timing of orders for GPU, HBM, cooling, or power equipment is off, related stocks will be sold in the short term.

However, if structural demand remains, the adjustment could become a preparation for 2027.

Risk 2｜Yen appreciation, interest rates, economic recession

Semiconductors and electronic components are affected by exchange rates and the global economy.

If the yen continues to appreciate, it will put downward pressure on companies with a high export ratio in the short term.

Additionally, if US interest rates remain high and the global economy slows, the pace of data center investment may slow.

However, for the domestic defense sector, the strong yen may lead to lower costs for some imported components.

Risk 3｜Delay in defense budget execution

Although the defense budget is a national policy, it takes time for it to be reflected in corporate performance.

This is because there is a flow of bidding, contracting, design, procurement, production, and acceptance inspection.

If the acceptance inspection is delayed, the sales expected for 2027 may be shifted to 2028 or later.

This is not a failure of the growth scenario;

“Postponing the period”

It is.

However, in the stock market, if the numbers do not come out at the expected time, stocks will be sold in the short term.

Investment Strategy｜Check the 2027 theme in the financial statements

The market for buying just the words AI, semiconductors, and defense is already coming to an end.

What is important for 2027 are the numbers that can be confirmed in the financial statements.

The numbers to look at are

  • Orders received
  • Order backlog *Operating profit margin *Segment profit *Capital investment
  • Research and development expenses
  • Medium-term plan target
  • EPS forecast

It is.

If the theme is genuine, it will definitely show up in the numbers.

On the other hand, stocks that do not show up in numbers just by their theme name will be easier to select in the 2027 market.

Final conclusion

Stock price fluctuations immediately after the announcement of financial results for May 2026 tend to be noise when viewed from a long-term theme.

The true value of the national policy triple theme of AI, semiconductors, and defense will be realized in the fiscal years ending March 2027 and March 2028.

“EPS expansion”

It depends on whether or not it can be proven.

What matters is not whether it's popular now.

The important thing is that

“Will the numbers change in 2027?”

It is.

In AI, power, cooling, post-processing, and highly reliable parts.

In semiconductors, HBM, advanced packaging, and edge AI.

In defense, space, cyber, electronic warfare, and communications networks.

In areas where these three overlap, there are candidates for performance transformation in 2027.

What investors need to do now is not get bogged down in short-term ups and downs.

Check order backlog, profit margin, medium-term plan, policy budget,

“A company where themes turn into profits”

It is to select.

The winners of 2027 are already starting to emerge in the backlog and investment plans for 2026.

Source

This article is for educational and informational purposes only, based on public information. It is not a recommendation or solicitation to buy or sell any specific security or financial product. Although care is taken with accuracy, the content and future investment outcomes are not guaranteed. Final investment decisions should be made at your own judgment and responsibility.