Ajinomoto ABF, Photoresists, and Helium: Three Vulnerabilities That Could Halt AI Semiconductors

If the straits in the Middle East close, AI stops: A chain reaction starting from a single film

Not many people would immediately imagine that a company called Ajinomoto holds the key to the global semiconductor industry. However, neither NVIDIA's cutting-edge AI chips, Intel's server CPUs, nor the computers you use every day could exist without a specific film made by Ajinomoto Fine-Techno.

And now, whispers in the corners of the industry suggest that this film 'might become impossible to produce.' Tracing the cause leads to a distant place in the Middle East: a narrow waterway called the Strait of Hormuz. It is a story of how a single thin thread connects seemingly unrelated events.

The heart of semiconductors made by a seasoning company

There is a material called ABF. The name is an acronym for Ajinomoto Build-up Film, and an Ajinomoto group company holds a 98% global market share. It is a thin insulating film sandwiched between semiconductor chips when they are mounted onto a substrate.

Why Ajinomoto? The story goes back to the 1970s, to research on byproducts generated during the production of monosodium glutamate. It was discovered that amino acid-derived resins could be applied to electronic materials, and after repeated research, it became the standard material for the semiconductor industry. It is a strange coincidence that byproduct research from a food company now supports a corner of the AI boom.

This film is used in large quantities, especially in cutting-edge AI semiconductors. While a standard PC CPU uses 6 layers of ABF, AI chip substrates like those from NVIDIA require 18 layers. Because the surface area of the substrate is also 3.5 times larger, the amount used per unit jumps nearly tenfold. The AI boom itself is explosively driving up demand for ABF.

The starting point of this article is the possibility that this ABF 'might become impossible to produce.'

The journey of raw materials: The film begins with Middle Eastern oil

Tracing where ABF comes from leads to an unexpected place.

Middle Eastern crude oil → naphtha obtained through distillation → benzene (aromatic compound) extracted from that → phenol → bisphenol A → epoxy resin → ABF. If you descend the steps of chemistry one by one, you end up with that film.

What is important here is that the amount of aromatic compounds—benzene and related substances—that can be obtained varies depending on the type of crude oil. Crude oils like Arabian Light from the Middle East are rich in aromatics. On the other hand, shale oil, which has seen increased production in the U.S. in recent years, is chemically 'light' and rich in paraffins (straight-chain hydrocarbons) but poor in aromatics. Even though they are both called 'crude oil,' they are chemically quite different materials.

Japan's petrochemical industry has been built consistently on the premise of Middle Eastern crude oil since the postwar period. Naphtha crackers, reformers, and the downstream benzene plants are all optimized with the 'composition of Middle Eastern crude oil' in mind. About 95% of the crude oil Japan imports comes from the Middle East, and most of it is transported through the narrow waterway known as the Strait of Hormuz.

And in the spring of 2026, an abnormal situation occurred in that waterway.

The narrow waterway of Hormuz

The Strait of Hormuz is a waterway about 34 km wide at its narrowest point, connecting the Persian Gulf and the Arabian Sea. Approximately 20% of the world's seaborne crude oil transport, and for Japan, over 90% of its imported crude oil, passes through this waterway. There is no other geographical singularity on Earth that holds the economies of so few countries in one single point.

Entering 2026, the passage through this strait has been effectively restricted due to tensions in the Middle East. Looking at Japan's trade statistics for March, naphtha imports from the Middle East were down 40% compared to the same period last year, and the operating rates of ethylene plants are also falling. While crude oil itself can be managed for the time being with reserves (254 days' worth), naphtha-derived chemicals are not subject to reserves, and if imports decrease, the impact is felt immediately.

Some might say, 'Then we should just buy American shale,' but this is where the chemical fact mentioned earlier comes into play. You cannot obtain enough aromatics from shale oil. It would take years and significant costs to modify Japanese refineries for shale, and even then, the aromatic yield would not match that of Middle Eastern crude. A bottleneck of quality, not quantity, arises: 'Even if we can secure crude oil, the aromatics decrease.'

The illusion that 'there is plenty in China'

Bisphenol A, an intermediate raw material for ABF, is actually in a state of global oversupply. China has continued large-scale expansion since the beginning of the 2020s, and the price of general-purpose products continues to fall. Japan's Mitsubishi Chemical shut down its production facilities in Kitakyushu in 2024, but this was not due to a lack of demand; it was because they lost the price competition against Chinese products.

So, if one asks whether we should just buy ABF raw materials from China, the answer is not that simple.

Bisphenol A for semiconductors is called "electronic grade" and is a different market from general-purpose products. While general-purpose products have a purity of 99.5%, electronic grade requires 99.95% or higher. Impurities like sodium and chlorine must be in the ppb (parts per billion) range rather than the ppm (parts per million) range of general-purpose products—meaning purity control 1,000 times stricter is required. The reason is simple: if trace amounts of metal ions enter a semiconductor chip, electricity will flow between the circuits and destroy them. When circuits become 5 nanometers or smaller, as in AI chips, even a single impurity can lower the yield.

There are only a handful of companies in the world that produce electronic-grade bisphenol A: Honshu Chemical Industry, Nippon Steel Chemical & Material in Japan, Chang Chun Petrochemical in Taiwan, and Kukdo Chemical in South Korea. These companies dominate the market, and all of them rely on naphtha derived from Middle Eastern crude oil for their raw material, benzene. No matter how much surplus China has in general-purpose products, it cannot be used for ABF.

Asia is in the same boat

"If Japan runs out, can't we get some from South Korea or Taiwan?"—It is natural to think that, but reality is not so kind.

South Korea also depends on the Middle East for 70% of its crude oil imports. Taiwan is on par with Japan. The electronic-grade bisphenol A produced by Taiwan's Chang Chun Petrochemical also comes from crude oil via the Strait of Hormuz. The petrochemical plants of South Korea's SK, LG Chem, and GS are also optimized for Middle Eastern crude oil. Only China has some independent procurement routes via pipelines from Russia (the ESPO pipeline), but even then, half of its supply comes via the Middle East.

In other words, the structure is such that if the Strait of Hormuz is blocked, the entire East Asian region will fall into a crisis simultaneously. It is not just Japan that will be in trouble; neighboring countries will be in the same situation. The idea that "we can just share resources within Asia" is an optimistic view based on peacetime distribution structures and will not hold up in a Hormuz blockade scenario.

If there is any salvation, it is that the unit price of electronic materials like ABF is orders of magnitude higher. If general-purpose polystyrene costs 200 yen per kilogram, electronic-grade intermediates cost from several thousand to tens of thousands of yen per kilogram. If aromatics become scarce, the production of general-purpose products will stop first, and the high-priced items for electronic materials will be secured until the end based on price—that is the structure. However, that is "securing electronic materials first" at the expense of the general chemical industry, and it does not change the fact that the Japanese chemical industry as a whole will suffer a major blow.

Actually, ABF was just the tip of the iceberg

I have been talking about ABF so far, but as you investigate the semiconductor material supply chain, you realize that ABF is just one of many vulnerable materials.

What might be even more serious is photoresist.

Photoresist is a photosensitive liquid used to draw fine circuits on semiconductors. It utilizes the property that the solubility of the exposed area changes to carve out circuit patterns. Japanese companies—Shin-Etsu Chemical, JSR, Tokyo Ohka Kogyo, Sumitomo Chemical, and Fujifilm—hold over 90% of the global market share. It is a daily consumable for the semiconductor industry, and unlike ABF, it is not a material that can be "stored up"; if it does not flow continuously, the manufacturing line will stop.

The components of that photoresist are also almost entirely specialty chemicals derived from naphtha. The base resin, solvents, additives, and the "photoacid generator (PAG)" that reacts to light to generate acid—all are made through the precision synthesis of aromatic compounds and fluorine compounds. It is an extremely fragile structure that cannot be manufactured if even one component is missing.

Three independent crises converging at one point

Photoresist is particularly serious because, in addition to a Hormuz blockade, two other independent crises are progressing simultaneously.

One is that the American company 3M completely withdrew from the business of fluorine compounds called "PFAS" at the end of 2025. PFAS are called "forever chemicals" because they do not decompose in the environment, and 3M announced its withdrawal in 2022 due to health damage and litigation risks. The problem is that 3M was also a major supplier of the specialty fluorine compounds that serve as the raw material for photoacid generators, the heart of photoresist. Since the withdrawal was announced in advance, the industry had a preparation period, but switching to alternative suppliers (Solvay in Belgium, AGC and Daikin Industries in Japan) takes time, including the certification process.

The other is the comprehensive PFAS regulation proposal being advanced by the European Chemicals Agency (ECHA). This was proposed in 2023 and aims to ban the use of PFAS within Europe in principle. The semiconductor industry is seeking an exemption on the grounds that there is "no alternative technology," but the timing of the regulation's entry into force and the scope of the exemption will reach a decision phase in 2026–2027.

In other words, for photoresist, three crises with completely different origins and natures—a Hormuz blockade (geopolitics), 3M's withdrawal (corporate decision), and ECHA regulations (environmental policy)—are progressing in parallel. And all three converge on a single point: photoresist.

You might think, "Why not just develop an alternative photoresist that doesn't use PFAS?" Research is underway all over the world. However, the current method, chemically amplified resist, is designed on the premise of the extreme acid strength of fluorine compounds—one trillion times that of ordinary acids—and cannot be easily replaced. Practical application requires a total of 5 to 10 years for development, formulation, and certification, and even in the industry's international roadmap, full-scale mass production introduction is slated for the 2030s or later. Over 99% of current photoresists still depend on fluorine compounds.

And, something even more fundamental—helium

I have discussed films, resists, and materials, but there is one more vulnerability that cannot be overlooked. That is the helium gas used to operate semiconductor manufacturing equipment.

Helium is the gas used to inflate balloons. However, in semiconductor manufacturing, it is used to cool wafers in etching equipment—the machines that carve circuits using plasma. While the wafer is heated by the heat from the plasma, the surface temperature must be managed in 0.2-degree increments. To achieve this, a thin layer of helium is flowed onto the back of the wafer to dissipate the heat.

The reason helium must be used is that its thermal conductivity is orders of magnitude higher than that of nitrogen or argon—six times higher than nitrogen and eight to nine times higher than argon—and it is chemically inert and has a low molecular weight, allowing it to penetrate tiny gaps. There is no other gas on Earth that possesses all of these properties.

Furthermore, helium is produced primarily in a limited number of countries, such as the United States, Qatar, Algeria, and Russia. It is extracted as a byproduct from natural gas fields, so production cannot be easily increased even if demand rises. If supply is interrupted, the manufacturing processes for the most high-value-added semiconductors—such as cutting-edge 3nm and 2nm logic, 3D NAND flash memory, DRAM, CMOS image sensors, and HBM (high-bandwidth memory for AI)—will immediately become impossible. Unlike photoresist or ABF, where the impact takes months to manifest, production would stop within the day.

What if everything happened at once?

ABF production decreases. Photoresist supply becomes restricted. Helium does not arrive. What happens if these occur simultaneously?

First, the production of AI semiconductors will stop. NVIDIA and AMD GPUs require ABF substrates, are patterned with photoresist, and go through etching processes cooled by helium. AI chips, which are currently seeing the fastest growth in demand, are the most dependent on all three of these materials. If production stops, data center investment plans by hyperscalers like Microsoft, Meta, Google, and Amazon will be delayed, and the growth of generative AI services themselves will slow down.

Next, server CPUs, PC CPUs, HBM memory, and high-performance chips for smartphones will be affected. Legacy semiconductors—mature chips for automotive and home appliance use—will be less affected, but the impact will be deeper the closer one gets to cutting-edge products.

From 2020 to 2022, the world experienced a semiconductor shortage. The pandemic, water shortages in Taiwan, and US-China tensions combined to reduce automobile production by tens of millions of units and drive up the prices of home appliances. The scenario currently being feared could be on a scale that exceeds that. This is because it would not be separate shocks for each material, but a chain reaction stemming from a common root: upstream Middle Eastern crude oil.

The obscure option of Russian oil

The most geographically and chemically rational alternative supply source is, in fact, Russian crude oil. Urals and ESPO crude oils have high aromatic yields and meet electronic-grade quality standards. Moreover, from Sakhalin in the Russian Far East or Eastern Siberia, it is transported via the ESPO pipeline to the port of Kozmino and can reach Japan in three to four days across the Pacific. It does not pass through the Strait of Hormuz at all.

However, since the invasion of Ukraine in 2022, Japan has effectively stopped direct imports of Russian crude oil due to the G7 price cap system. The only exception is the Sakhalin-2 project (in which Mitsui & Co. and Mitsubishi Corporation hold interests), and even this is limited.

What is interesting is that products derived from Russian crude oil may be reaching Japan via other routes. India purchases large quantities of Russian crude oil at a discount, refines it at facilities like the Jamnagar refinery, and exports naphtha and aromatic products to the world. These are flowing into Japan indirectly. While it is not recognized as "Russian-made" at the crude oil stage, it is present at the product stage—a structure known in the industry, though no one speaks of it openly.

If the blockade of the Strait of Hormuz is prolonged, political compromises such as the expansion of Sakhalin-2 or a de facto easing of the price cap will become realistic. During the first oil crisis in 1973, Japan had a precedent of shifting its diplomatic policy toward the Arab nations. Energy security ultimately forces political choices.

What was invisible becomes visible

Usually, we do not think about what is inside our smartphones. When we ask AI a question, we are not conscious of the fact that NVIDIA chips are performing calculations behind the scenes. The film that fixes those chips to the substrate is made by an affiliate of a food company called Ajinomoto, and its raw materials are transported from Middle Eastern oil fields through a narrow strait—it is easier not to think about such things if you can avoid it.

However, the entire "system that allows us not to think" is actually dependent on a surprisingly small number of locations, companies, and substances. Almost all of Japan's ABF is produced at Ajinomoto Fine-Techno's Gunma plant. Only a few companies in the world can produce electronic-grade bisphenol A. 90% of photoresist is produced by Japanese companies. There are only a handful of countries that can stably supply helium. The transport of Middle Eastern crude oil hinges on the Strait of Hormuz.

If one point stops, others stop in a chain reaction. That is the structure that lies beneath the modern semiconductor industry.

This is not a story about "Japan being bad" or "the Middle East being bad." Since the latter half of the 20th century, the world has pursued division of labor and efficiency. It is cheaper for one company to supply one material to the entire world. But because of that, the impact when that one point stops becomes greater. The modern premise of "cheap, high-quality semiconductors" that we enjoy is built upon this concentration.

If the situation in the Strait of Hormuz is resolved early, many scenarios will end up being groundless fears. However, the 3M withdrawal and ECHA regulations are independent variable factors that are progressing regardless of the Strait of Hormuz, and they will continue to shake the supply structure of semiconductor materials over the next 5 to 10 years. Even if the Strait of Hormuz issue is resolved, the structural vulnerability itself will not disappear.

The story that began with Ajinomoto's film has traveled unexpectedly far. Yet, this "unexpected distance" is the true nature of modern industry. To produce a single sheet of film, a supply chain spanning half the globe is set in motion. And if any single point along that path becomes blocked, state-of-the-art AI comes to a halt.

Things that were once invisible are now beginning to come into view, little by little, through the narrow waterway of Hormuz.

This report is based on public information, industry analysis, and corporate IR materials confirmed as of May 2026. Specific figures and timing projections contain uncertainties, and there are many factors invisible from the outside, such as the duration of the situation in the Strait of Hormuz, the inventory levels of individual companies, and the actual supply capacity of alternative suppliers. This report is a compilation of information intended to help understand industrial structure and does not constitute investment advice for specific companies or an advocacy for any particular policy.