The "Hidden Hero" of AI Semiconductors is Inspection Equipment - Japan's Advantest Holds the Key to the Next Growth Market

When it comes to companies symbolizing the generative AI boom, many investors would first think of NVIDIA.

Surrounding it are companies like TSMC, SK hynix, Micron Technology, and Broadcom, and in semiconductor manufacturing equipment, there are companies like ASML and Tokyo Electron.

However, if you trace the AI semiconductor supply chain a little further, a crucial process that has been relatively unnoticed by general investors emerges.

That is "testing."

Semiconductors do not become products just by being manufactured.

Do they operate as designed? Do they avoid errors during high-speed calculations? Do they function normally within the expected voltage and temperature ranges? Can defective products be reliably eliminated from mass-produced chips?

Only after these confirmations can expensive AI chips be sent to customers.

An article by Aktiencheck, which reprinted an article from the German investment media DER AKTIONÄR, also highlighted this "less visible part of the AI supply chain."

And there is a particularly interesting reason for Japanese investors regarding this theme.

One of the companies with overwhelming presence in the global semiconductor test equipment market is Japan's Advantest.

As AI chips become more expensive, the cost of "failing and then discarding" also rises.

To understand the importance of semiconductor testing, it's necessary to consider that current AI chips are becoming quite different products from traditional semiconductors.

The latest AI accelerators are not simple single semiconductors.

They typically have a structure where HBM (High Bandwidth Memory) is placed around GPUs or dedicated AI processors, combining multiple semiconductors into one massive computing system using advanced packaging technology.

In other words, the number of components in a single product increases, connection points increase, power consumption rises, and the signal speed handled also increases.

Naturally, the number of inspection items also increases.

What is important here is that the more expensive the product becomes, the greater the loss when a defect is found after completion.

For inexpensive semiconductors, simply discarding defective products may suffice, but for AI packages combining multiple expensive chips and HBM, the situation changes.

If a problem in just one component renders the entire finished product unusable, it could waste materials worth thousands of dollars, or even more in some cases.

That is why manufacturers strive to identify "normal chips" at an earlier stage of the manufacturing process.

As AI semiconductors become more advanced, the testing process does not diminish; rather, it increases.

The magnitude of "AI test demand" indicated by Advantest's figures

Advantest's performance clearly illustrates this structural change.

In the first quarter of fiscal 2026, announced on July 29, 2026, covering April to June 2026, sales were 367.5 billion yen.

This was a 39.3% increase compared to the same period the previous year.

Operating profit was 190 billion yen, a 53.3% increase from the same period the previous year.

Net profit was 174.8 billion yen, a 93.8% increase from the same period the previous year.

Rather than a simple recovery in the semiconductor market, the expansion of demand for high-performance semiconductor testers for AI is significant.

The company also explains that AI-related HPC semiconductors and high-performance memory for data centers are driving the semiconductor market, with increased production and device complexity leading to active tester investments.

Furthermore, the company significantly raised its performance forecast for the fiscal year ending March 2027.

The sales forecast is 1.714 trillion yen.

This represents an approximately 52% increase compared to the previous fiscal year.

The operating profit forecast was also raised to 846 billion yen, expecting an approximately 70% increase from the previous fiscal year.

While there are companies worldwide with "AI" in their name, not many reflect it as clearly in their actual profits.

This is why it is difficult to dismiss the current Advantest as merely an "AI-related stock."

Rival Teradyne also hits record highs—indicating market expansion

Interestingly, it's not just Advantest that is performing well.

The largest competitor, U.S.-based Teradyne, also recorded near-record performance in the second quarter of 2026.

Sales were $1.329 billion.

This more than doubled from $652 million in the same period the previous year.

Of this, the semiconductor test division alone accounted for $1.122 billion.

The company explains that AI-related demand continues to be strong not only in computing semiconductors but also in the memory sector.

Test demand for HBM and DRAM is particularly strong, with memory test sales exceeding $200 million for three consecutive quarters.

In other words, it's not a simple scenario of "Advantest growing by taking market share from Teradyne."

At this stage, it seems more natural to think that the AI semiconductor market's growth is significantly expanding the test market itself.

As investments in AI data centers continue, and as custom ASICs, HBM, network semiconductors, and optical communication devices increase, quality assurance becomes necessary for all of them.

An increase in NVIDIA's sales benefits not only GPU manufacturers.

TSMC's production volume increases.

HBM manufacturers increase production.

Packaging demand increases.

And ultimately, the number of semiconductors that need to be tested also increases.

Semiconductor test companies are getting closer to a position where they can charge a "toll" on AI capital investments.

From Japan's perspective, an even more interesting supply chain emerges

What is important for Japanese investors is that the theme of semiconductor testing does not conclude with just one company, Advantest.

In the testing process, numerous peripheral devices are required, such as probe cards and test sockets to connect semiconductors to inspection equipment, probers for precise wafer positioning, and temperature control devices.

In other words, as AI semiconductors become more complex, the benefits may extend to surrounding equipment and components.

In June 2026, Teradyne announced an integrated test solution for AI and data center devices with Tokyo Electron.

Despite being competitors, Japanese companies appear in the ecosystem supporting their products.

This is the characteristic of the current semiconductor industry.

It is not an industry neatly divided by nationality.

U.S. companies design, Taiwanese companies manufacture, Japanese and U.S. companies inspect with equipment, combine with Korean company memory, and ship to data centers worldwide.

In this context, Japanese companies secure multiple positions where they may not always stand out as brand names of finished products but are essential for semiconductor production.

This is also why Japanese companies remain strong in semiconductor materials, manufacturing equipment, inspection equipment, and precision components.

Why investors searching for "the next NVIDIA" are turning to inspection equipment

In the early stages of the AI market, investors' funds first concentrated on NVIDIA.

Subsequently, it expanded to memory-related companies like SK hynix and Micron, driven by HBM demand.

The investment theme is further expanding to power equipment, data center cooling, optical communication, and network equipment.

This is a common movement in the stock market.

When a major technological innovation occurs, the initially obvious leading companies are bought, and then funds move to what is absolutely necessary for those leading companies to grow.

Semiconductor testing is likely one of those.

What makes the test market particularly interesting is the possibility that the complexity of AI chips directly leads to market expansion.

It's not just about increasing the number of chips produced.

The time to inspect a single chip increases.

The number of inspection items increases.

The number of tests conducted, both during manufacturing and after completion, increases.

Moreover, AI semiconductors are extremely expensive.

If a chip with insufficient performance or potential failure is incorporated into a data center, it incurs not only replacement costs but also losses from server downtime.

As a result, the amount that can be paid for quality assurance increases.

The important point in this market is not just "the number of semiconductors × test unit price," but "the amount of testing per semiconductor itself increases."

On social media, opinions are split between optimists and skeptics

However, when considering stock investment, the situation becomes a bit more complex.

When looking at the reaction of Japanese investors to Advantest, there is a significant temperature difference.

On the Advantest board on Yahoo! Finance, as of August 24, about 42% of recent posts in the past week were in favor of "strongly want to buy" and "want to buy."

On the other hand, about 33% expressed "want to sell" and "strongly want to sell."

In other words, it is not overwhelmingly optimistic.

Posts from optimists include thoughts like "AI demand is still in its early stages," "there is performance backing as a niche leader," and "as long as AI semiconductor demand continues, test demand will also grow."

Meanwhile, skeptics voice concerns such as "the stock price has already risen significantly," "there is a high risk of entering at high prices," and "one should be cautious after a securities firm's target price increase."

As of August 21, Advantest's stock was at 35,900 yen.

On August 14, it recorded a year-to-date high of 38,730 yen.

The company's forecasted PER has also risen to about 39 times.

Strong performance and a stock being undervalued do not mean the same thing.

This is the main reason why opinions are divided on social media.

Overseas investors are also searching for "the next bottleneck in AI"

The same trend is observed in overseas investment communities.

On Reddit, posts about Teradyne evaluate the increase in test processes due to the complexity of HBM and ASIC as investment material.

Additionally, discussions are increasing about how the overall AI infrastructure stocks are no longer being evaluated just because they are simply AI-related, despite continued AI demand.

This is an important change.

In the AI market from around 2023 to 2024, there were instances where a company's evaluation rose just because it was related to AI.

However, as the market matures, the points investors check change.

Is revenue really increasing?

Is the operating profit margin improving?

Is there a competitive advantage?

Will customers continue their capital investments?

Will profit growth continue to justify high stock valuations?

Advantest and Teradyne, at least at this point, are demonstrating demand expansion with actual financial figures, not just the word "AI."

In that sense, they fall into a relatively easy-to-understand category among the numerous AI-related stocks.

On the other hand, the biggest risk is also "AI"

Ironically, AI, the biggest growth factor for semiconductor test companies, can also become the biggest risk.

Currently, global tech giants like Google, Microsoft, Amazon, and Meta are investing massive amounts in AI infrastructure.

As long as this investment continues, the production of high-performance semiconductors and HBM will increase, and test demand will also easily expand.

However, if investment in AI data centers reaches a peak, capital investment could rapidly decelerate.

The semiconductor manufacturing equipment market historically experiences significant economic cycles.

When demand is strong, customers place large orders for equipment, but once factories are sufficiently equipped, new orders can sharply decline.

Furthermore, test companies have a high dependency on a few large customers.

Teradyne itself explains in its SEC filings that a few key customers account for a large portion of demand in the test product market.

Even if AI demand continues, quarterly performance can fluctuate significantly depending on the timing of customer companies' capital investments.

For high-growth companies, even a slight decline in growth rate can lead to significant stock price adjustments.

For stocks like Advantest, which are expected to grow significantly by the market, what matters more than whether the earnings are "good or bad" is whether they were "as good as the market expected."

What is more important than deciding on a single "winner"##HTML_TAG_268