Kioxia, Advantest, Lasertec.
I believe many people hold or are watching these three stocks.
They are the protagonists of the Japanese semiconductor market.
However, when certain signs begin to appear, you should reconsider your positions in these three stocks.
That is when market capital begins to shift to 'Light.'
'Light' refers to photo-electronic convergence (Silicon Photonics).
It is the theme of the next semiconductor cycle and is just beginning to take off now.
Why does the semiconductor era end when 'Light' arrives?
What is Driving the Current Semiconductor Cycle?
First, let's confirm the structure driving the current semiconductor market.
Meta, Microsoft, Google, Amazon, Apple.
These five Big Tech companies are making massive capital investments in data centers.
The breakdown of that capital investment is as follows:
- Buying massive amounts of computing chips (NVIDIA)
- Buying manufacturing equipment to make chips (Tokyo Electron)
- Buying inspection equipment for chips (Lasertec, Advantest)
- Buying massive amounts of memory to store data (Kioxia)
The capital investment of these five Big Tech companies is driving the entire semiconductor cycle.
The crucial perspective here is one I have mentioned before:
The payment capacity of these five companies determines the ceiling for the semiconductor industry's growth rate.
The industry's growth rate is not infinite. It is determined by the total payment capacity of the buyers.
So, What is 'Light'?
Current data centers have a fatal problem.
Power.
Computing chips perform massive calculations.
The results of those calculations are sent from chip to chip and from server to server.
Currently, copper wiring (electrical signals) is used for this 'sending' part.
The problem is that copper wiring generates heat when electricity passes through it.
In large-scale data centers, about half of the network-related costs and power consumption are caused by this data transmission (copper wiring).
Power is being wasted on moving data between chips rather than on the calculations themselves.
This is the 'Power Wall.'
Photo-electronic convergence is the technology that replaces this copper wiring with optical fiber.
Light generates almost no heat.
According to NVIDIA's estimates, switching to light can reduce power consumption by up to 70%.
In other words, 'Light' will break the biggest wall that data centers are currently hitting.
NVIDIA is Already Betting on 'Light'
This is a decisively important point.
In March 2026, NVIDIA invested in two U.S. companies with photo-electronic convergence technology.
- Coherent Corp: Approx. 310 billion yen
- Lumentum Holdings: Approx. 310 billion yen
- Total: Approx. 6200 billion yen
Furthermore, they signed parts purchase agreements worth hundreds of billions of yen.
They officially announced next-generation optical connection technology and declared that light would be placed at the core of data centers.
NVIDIA is a company that sells its own computing chips. That same NVIDIA has bet 620 billion yen on the 'outside of the chip.'
What does this mean?
'There is a limit to just increasing chip performance. The next bottleneck is how they are connected.'
NVIDIA has made this judgment itself.
In other words, the transition from a phase of increasing the quantity of semiconductors to a phase of changing how they are connected has begun.
Why Does the Semiconductor Era 'End' When Light Arrives?
This is the main point.
Technology has cycles. When investment concentrates on a certain theme and it becomes saturated, capital moves to the next theme.
The beneficiaries of the current semiconductor cycle are...
- Kioxia (Increased memory production)
- Advantest (Chip testing equipment)
- Lasertec (Mask inspection equipment)
These three companies are beneficiaries of the 'increasing chip volume' phase.
Make chips in large quantities
-> Memory sells
-> Testing equipment sells
-> Inspection equipment sells.
However, when photo-electronic convergence becomes full-scale, the investment destination for data centers will change.
It will change from 'increasing chip volume' to 'changing chip connectivity.'
When the axis of investment shifts, capital also shifts.
Capital that was previously headed toward Kioxia or Advantest will begin to move toward photo-electronic convergence companies. This is the structure of 'When Light arrives, the semiconductor era ends.'
It's not that semiconductors themselves become unnecessary. However, the growth rate of the semiconductor cycle will peak, and the priority of capital will change. The moment growth slows down, stock prices fall much further than the actual situation warrants.
The fact that Kioxia's stock price has fallen 56% from its high despite an operating profit of 1.27 trillion yen is exactly due to this structure.
Who are the Beneficiaries of 'Light'?
Let's look at the beneficiaries of the next cycle.
However, this is not a 'buy now' list. Cycle transitions happen in stages.
Companies making optical components:
- Coherent Corp (USA): Approx. 310 billion yen investment from NVIDIA
- Lumentum Holdings (USA): Also approx. 310 billion yen investment from NVIDIA
- Suntech Holdings (6777): A global presence in tunable light sources and optical measuring instruments. For the fiscal year ending March 2026, sales were 31.5 billion yen, operating profit was 10.3 billion yen, with a profit margin exceeding 30%.
- Furukawa Electric (5801): A veteran in optical fibers and optical devices.
- Sumitomo Electric Industries (5802): A global leader in optical fibers.
Companies making optical chips:
- QD Laser (6613): Quantum dot laser technology. A core device for silicon photonics. It is in the red, but the technology is unique.
- Hamamatsu Photonics (6965): A world leader in optical sensors and detectors.
Supporting players (Inspection, Packaging, Materials):
- Dexerials (4980): Optical films, precision bonding materials. Significantly revised its medium-term plan upward. Demand is surging in the packaging process for photo-electronic convergence.
- Resonac (4004): Advanced packaging materials.
- Ushio Inc. (6925): Light source technology.
Optical Platforms:
- NTT (9432): Promoting photo-electronic convergence technology through the 'IOWN' initiative. Scheduled to start commercial provision of photo-electronic convergence switches during FY2026.
Points to Note Here
Note 1: This is not 'Sell semiconductors immediately'
The mass production and commercialization of photo-electronic convergence will proceed in stages from 2026 to 2028. This is not a story for today or tomorrow. It's not that the semiconductor cycle 'ends today,' but that it is 'approaching its peak.' What you should watch is the 'pace of capital inflow into light-related stocks.' When this begins to accelerate, it's time to review your semiconductor positions.
Note 2: Many photo-electronic convergence stocks are still in the 'expectation' stage
QD Laser is in the red. Suntech is profitable and has a track record, but its scale is small. There are many stocks for which it is still too early to buy with full force based on the conviction that 'Light is coming.' In cycle transition periods, the timing to 'sell the winners of the current cycle' and 'buy the winners of the next cycle' often diverges. You will get burned if you jump the gun.
Note 3: Photo-electronic convergence changes semiconductors rather than killing them
Photo-electronic convergence is a technology that integrates optical devices into chips. Semiconductors will not disappear. 'The semiconductor era ends' means 'the beneficiary structure of the current semiconductor cycle ends.' Semiconductors themselves will continue to be used in the next cycle. However, the faces of the companies taking the profits will change.
Perspectives Investors Should Have
Just three simple points:
1. Trading volume and capital inflow of light-related stocks
Suntech, QD Laser, Dexerials, Furukawa Electric. If the trading volume of these stocks begins to surge, it's a signal that the capital shift has started.
2. Changes in the content of Big Tech capital investment
In the financial results of Meta, Microsoft, and Google, if the breakdown of capital investment begins to shift from 'purchasing chips' to 'developing optical infrastructure,' it's a confirmation of the cycle transition.
3. NVIDIA's next move
Whether NVIDIA makes further additional investments in photo-electronic convergence or moves to produce optical devices in-house. NVIDIA's actions will be the most reliable leading indicator of the cycle transition.
Summary
The end of a semiconductor cycle does not necessarily begin with a crash.
In many cases, it begins with 'capital quietly starting to move to the adjacent theme.'
Right now, that 'adjacent theme' is photo-electronic convergence.
NVIDIA bet 620 billion yen.
NTT is starting commercialization with the IOWN initiative.
Suntech's operating profit margin exceeded 30%, hitting record highs.
Many investors haven't noticed yet. While they are fretting over the financial results of semiconductor stocks, light-related stocks are quietly starting to move.
Kioxia, Advantest, Lasertec.
Those who hold these should not just look at the financial figures.
You should simultaneously watch 'whether capital is moving to Light.'
When Light arrives, the semiconductor era ends.
'Ends' is not today.
However,
the results will differ greatly between investors who missed that signal and those who were watching it.
Cycle transitions always begin quietly.
