Does Japan have its own major AI model company?

Not at the scale of OpenAI, Anthropic, or Google DeepMind. Japan's public-market AI exposure is overwhelmingly on the supply side — chipmaking equipment, memory, and infrastructure — rather than foundation models themselves.

What does Tokyo Electron actually make?

Semiconductor fabrication equipment used to etch, deposit, and coat the silicon wafers that become chips. It's one of the largest chip-equipment makers in the world, alongside US and Dutch peers.

Why does memory (Kioxia) matter for AI?

Large AI models and the servers that run them need enormous amounts of fast storage and memory, not just compute. Kioxia is one of the world's major NAND flash memory makers, benefiting from that demand alongside the DRAM/HBM cycle led by Korean and US producers.

What's the biggest risk specific to this sector?

Concentration. A relatively small list of names captures most of the AI-linked trade in Japan, so they tend to move together — a slowdown in global AI infrastructure spending would likely hit this group broadly rather than selectively.

IVYXON / Japan Stocks Guide / AI Sector Map

JAPAN STOCKS GUIDE · AI SECTOR MAP

Where Japan Fits Into the Global AI Buildout

Japan isn't home to a major AI model company or hyperscaler — there's no domestic equivalent of OpenAI, Anthropic, or the cloud giants tracked on our Hyperscaler CapEx Tracker. What it does have is a concentrated position further up the supply chain: the equipment that fabricates advanced chips, a large share of the world's memory, and a slice of the power infrastructure that AI data centers run on. This page maps out who does what.

Semiconductor Equipment

Before a chip design becomes a working piece of silicon, it passes through a long chain of specialized machines — and Japan makes several of the most important ones.

TOELY

Tokyo Electron

One of the world's largest chip-fabrication equipment makers, alongside US and Dutch peers — tools for etching, deposition, and coating silicon wafers.

ATEYY

Advantest

Automated test equipment for chips, with a particularly strong position in memory testing — a step that scales directly with AI chip production volumes.

DSCSY

Disco Corp

Dicing and grinding equipment that cuts and thins silicon wafers — a specialized niche that becomes more important as advanced packaging (stacking multiple chips) grows.

LSRCY

Lasertec

Mask and wafer inspection equipment, with an unusually strong position in inspecting the photomasks used in EUV lithography — a bottleneck step for the most advanced chips.

—

Screen Holdings (7735.T, no US ADR found)

Wafer cleaning equipment — Tokyo-listed only; reachable through a broker with direct Tokyo Stock Exchange access or a Japan ETF, not a US ADR.

Memory

AI servers need enormous amounts of fast storage and memory, not just raw compute. Japan's biggest name here is Kioxia Holdings (KXIAY) — one of the world's major NAND flash memory makers, spun out of Toshiba's old memory business. Kioxia's own path to the public market was rocky: a planned 2020 listing was shelved amid US-China trade tensions, a proposed 2021 merger with Western Digital fell apart, and it finally debuted on the Tokyo Stock Exchange's Prime Market in December 2024. It's since become one of the more actively traded AI-linked names in Tokyo, riding the same storage and memory demand cycle that Korean and US producers are also benefiting from.

Power & Grid Infrastructure

The part of the AI story that gets less attention: data centers at this scale consume enormous amounts of electricity, and someone has to build the cables, transformers, and grid equipment to deliver it. Two Japanese industrial names sit in that layer:

FUWAY

Furukawa Electric

Optical fiber and power cables — the physical links that connect data centers to each other and to the grid.

MIELY

Mitsubishi Electric

Power semiconductors and grid/transmission equipment — the hardware layer behind delivering data-center-scale power reliably.

Diversified Tech & Services

A second tier of large, diversified Japanese companies has meaningful but less singular AI exposure — worth knowing about, even if AI is one growth driver among several rather than their whole story.

SFTBY

SoftBank Group

Holds a large stake in Arm, whose chip architecture licensing sits underneath much of the global AI chip industry, plus direct AI infrastructure investments.

FJTSY

Fujitsu

Enterprise IT services and hardware, increasingly positioning around domestic AI and data-center build-out.

NECPY

NEC

IT services, networking, and a long-running AI research arm — another domestic-infrastructure-adjacent play.

HTHIY

Hitachi

Broad industrial conglomerate with growing exposure to data-infrastructure and grid-modernization services.

RNECY

Renesas Electronics

Automotive and industrial microcontrollers and analog chips — positioned around AI moving to the "edge," not just the data center.

Full company names, listing details, and how to actually buy each of these are on the access guide.

⚠ The Risk Specific to This Group

Concentration. A relatively short list of names captures most of the AI-linked trade in Tokyo. When the theme is in favor, that's exactly what makes them liquid; when it isn't, they tend to fall together rather than independently.
They're downstream of US hyperscaler spending decisions. A pullback in the capital expenditure tracked on our Hyperscaler CapEx Tracker would likely flow through to Japanese equipment and memory names with a lag, not immunity.
Cyclicality is normal in this industry, independent of AI — semiconductor equipment and memory have always been boom-bust businesses. "AI-driven" doesn't mean "acyclical."
Currency. Most of these trade as unsponsored OTC ADRs or Tokyo-listed shares — see the access guide for what that means for spreads and currency exposure.

Continue Reading

This page provides general information about publicly traded companies compiled from public sources. It is not investment advice, and nothing here is a recommendation to buy or sell any specific security. Business descriptions are simplified and may not capture a company's full operations; company exposure to AI-related demand can change quickly. Verify anything material before acting on it, and consult a licensed professional for advice specific to your situation.