Japan AI Market 2026: A Complete Overview

Japan’s domestic AI systems market reached ¥1.34 trillion (roughly $9 billion) in 2024, growing 56.5% year on year, and IDC Japan forecasts it will more than triple to ¥4.19 trillion by 2029 at a 25.6% compound annual growth rate (IDC Japan, May 2025). That makes Japan one of the largest single AI markets in Asia — and one of the strangest. The money is enormous: hyperscalers including Amazon Web Services, Oracle and Microsoft have committed over $26 billion to Japanese data centres, and the government nearly quadrupled its chip and AI budget to ¥1.23 trillion for fiscal 2026. The usage is not. Only 34.9% of Japanese consumers use generative AI at least several times a month, versus 86.3% in China and roughly 60% in the US and Germany (Nomura Research Institute, September 2025 survey).

That gap — world-class capital deployment, laggard everyday adoption — is the defining feature of the Japan AI story in 2026, and the reason SoftBank, OpenAI, Anthropic, NTT and Sakana AI are all attacking the same prize: the enterprise middle. This is the market deep-dive companion to our access-focused guide to which AI models you can use in Japan, and a sibling of our Japan gaming market analysis.

How big is Japan’s AI market in 2026?

Start with the headline number: IDC Japan sized the domestic AI systems market at ¥1.34 trillion in 2024 and projects ¥4.19 trillion by 2029 — 3.1 times growth in five years (IDC Japan, May 2025). The infrastructure layer beneath it is moving even faster. Japan’s AI infrastructure spending expanded seven-fold between 2022 and 2025 and will surge past $5.5 billion in 2026, up 18% year on year, with a 13% CAGR through 2029 (IDC, March 2026). IDC expects 2028 to be the tipping point when AI infrastructure spending overtakes non-AI infrastructure spending in Japan for the first time.

The services layer is compounding too: IDC projects Japan’s AI services market will hit $8 billion by 2030, 3.2 times its 2025 level. For how Japan stacks up against China, India and Southeast Asia, see our Asia AI market country breakdown — Japan is the region’s clearest case of infrastructure running ahead of usage, which is precisely why vendors are pricing in a catch-up.

Why is Japanese AI adoption so cautious?

Japan’s adoption numbers remain the softest of any major AI market. NRI’s four-country survey conducted in September 2025 found 34.9% of Japanese respondents use generative AI at least several times monthly, against 86.3% in China and around 60% in both the US and Germany. Willingness to pay is weaker still: just 15.5% of Japanese users subscribe to paid AI services, versus 66% in China and 40.2% in the US (NRI, April 2026).

But the texture matters. NRI’s diagnosis is that Japan’s problem is “not a lack of trust, but a lack of sufficient impetus for use” — consumers trust the technology but feel no urgency, making Japan a rich, untapped market rather than a hostile one. The trend line supports that read: 42.5% of Japanese internet users had tried generative AI by February 2025, up nine percentage points in a single year, and active workplace usage rose from 15.7% to 19.2% between August 2024 and February 2025 (GMO Research, February 2025). ChatGPT dominates with a 54.9% share among active Japanese users, ahead of Gemini at 29.7% and Microsoft Copilot at 14.6% (GMO Research, 2025). Japan is not rejecting AI. It is adopting it at the pace of a country with lifetime employment, risk-averse middle management and an acute shortage of software engineers — which is exactly why the enterprise sales motion, not the consumer app, is where the money is.

Which Japanese AI companies matter?

The incumbents: NTT, Fujitsu, NEC

Japan’s telecoms and IT giants are betting on sovereignty and efficiency rather than frontier scale. NTT launched tsuzumi 2, its “purely domestic” lightweight LLM, in October 2025, pitching high security and low cost for corporate digital transformation, and by May 2026 was claiming world-class performance on Japanese business documents — charts and tables included — running on a single GPU (NTT press releases, October 2025 and May 2026). Fujitsu is pushing its Takane Japanese LLM into private, on-premises enterprise deployments, while NEC’s cotomi model targets the same data-sovereignty-conscious buyers. The pitch is consistent: Japanese-language performance, domestic infrastructure, no data leaving the country.

The startup: Sakana AI is Japan’s most valuable unicorn

Tokyo-based Sakana AI raised a $135 million Series B at a $2.65 billion valuation in November 2025 to build Japan-optimised AI models (TechCrunch, November 2025), making it Japan’s most valuable startup. Founded by former Google researchers, Sakana’s evolutionary model-merging approach is a deliberate counter-position to brute-force scaling — a very Japanese thesis that efficiency beats size. Preferred Networks, the deep-learning veteran behind the MN-Core accelerator line, rounds out a domestic ecosystem that our access guide counts at more than 30 homegrown LLM variants.

The Americans: SoftBank–OpenAI, Anthropic, Google

The most consequential structure in the market is SB OAI Japan, the 50/50 joint venture between OpenAI and SoftBank-controlled C Holdings launched on 5 November 2025 to sell “Crystal Intelligence”, an enterprise AI package combining OpenAI models with localised system integration, with availability from 2026 (SoftBank Group, November 2025). SoftBank is its own first customer — the group has already built roughly 2.5 million custom GPTs internally — and by February 2026 the JV was layering OpenAI’s Frontier enterprise platform into the offer (SoftBank, February 2026). Anthropic opened its first Asia-Pacific office in Tokyo in late October 2025, signing a Memorandum of Cooperation with the Japan AI Safety Institute and naming Rakuten, Panasonic and Nomura Research Institute as customers; the company says Japan ranks in the top 25% of countries on its Economic Index for AI adoption and its Asia-Pacific run-rate revenue grew more than 10x in a year (Anthropic, October 2025). The enterprise land-grab is on, and Japan’s cautious adopters are the territory.

The state is writing very large cheques

Tokyo has decided AI and chips are national security policy. The FY2026 budget approved in December 2025 nearly quadruples METI’s chip and AI spending to roughly ¥1.23 trillion ($7.9 billion), including ¥150 billion for 2nm chip venture Rapidus and ¥387.3 billion for AI — covering domestic foundation models, data infrastructure and “physical AI” that controls robots and machinery (Bloomberg via Taipei Times, December 2025). That sits inside a standing commitment of more than ¥10 trillion in public support for AI and semiconductors by fiscal 2030, designed to induce over ¥50 trillion in public-private investment across a decade — and the Takaichi government is now discussing a broader ¥60 trillion strategic-technology spending target for fiscal 2026–2030, double the previous five-year plan (CC Press, 2026).

The most distinctive bet is robotic. Under a METI/NEDO programme worth up to ¥1 trillion over five years, a new company called Noetra — majority-owned by SoftBank, NEC, Sony Group and Honda, with researchers from Preferred Networks and AIST — is building a national AI foundation model for robots, targeting 10 million AI-powered robots across 18 industries by 2040 (AI News, July 2026). In a country whose workforce is shrinking and whose migration policy remains tight, physical AI is not a moonshot; it is labour policy. No other major AI market has fused industrial strategy and demographics this explicitly.

Hyperscalers have committed $26 billion — and hit a power wall

Foreign cloud capital is arriving faster than Japan’s grid can absorb it. AWS has committed $15.5 billion to Japanese infrastructure, Oracle $8 billion for GPU-heavy expansion announced in December 2024, and Microsoft $2.9 billion across three regions — over $26 billion in confirmed hyperscaler commitments (Introl analysis, January 2026). The constraint is electricity: grid connection waits around Tokyo now run five to ten years, and Japanese data-centre power demand is projected to roughly triple from 19 TWh in 2024 to 57–66 TWh by 2034 — equivalent to the consumption of 15–18 million households (Introl, January 2026). That is pushing builds towards Hokkaido and Kyushu, and it is why Japan’s data-centre boom will be rationed by megawatts, not money.

Regulation is soft law by design

Japan passed its first AI law, the AI Promotion Act, in 2025 — and it is the anti-EU AI Act: a deliberately “innovation-first” framework focused on promoting research, development and utilisation, with no monetary penalties attached (White & Case, 2025). Obligations on business are cooperative rather than punitive, and enforcement runs through guidance and reputational pressure rather than fines. For AI vendors, Japan is now arguably the most permissive major developed market to operate in. Track how that compares across the region in our Asia AI policy tracker.

FAQ: Japan AI market 2026

How big is Japan’s AI market?

Japan’s domestic AI systems market was worth ¥1.34 trillion in 2024, growing 56.5% year on year, and IDC Japan forecasts it will reach ¥4.19 trillion by 2029. AI infrastructure spending alone will exceed $5.5 billion in 2026 (IDC, March 2026).

What percentage of Japanese people use generative AI?

As of September 2025, 34.9% of Japanese consumers used generative AI at least several times a month — far below China’s 86.3% and roughly 60% in the US and Germany (NRI). Around 42.5% of internet users had tried it at least once (GMO Research, February 2025).

Who are the biggest Japanese AI companies?

NTT (tsuzumi 2), Fujitsu (Takane), NEC (cotomi) and Preferred Networks lead the incumbents; Sakana AI, valued at $2.65 billion after its November 2025 Series B, is Japan’s most valuable AI startup. SB OAI Japan — the SoftBank–OpenAI joint venture — is the biggest enterprise distribution play.

How much is the Japanese government spending on AI?

The FY2026 budget nearly quadruples chip and AI support to about ¥1.23 trillion ($7.9 billion), within a pledge of over ¥10 trillion in public support for AI and semiconductors by fiscal 2030 (Bloomberg via Taipei Times, December 2025). A separate ¥1 trillion, five-year programme funds a national AI model for robots.

Is AI regulated in Japan?

Lightly. The 2025 AI Promotion Act is an innovation-first framework with no monetary penalties, relying on voluntary cooperation and government guidance rather than EU-style enforcement (White & Case, 2025).