Necessities · Markets · Japan

Why Japan.

6 min read JP

The window

Japan. In-flight regulations by year.

2024 0 2025 0 2026 0 2027 5 2028 6 2029 0 2030 0 Peak crystallisation 2028. 6 regulations

Each bar counts effective dates in Japan between 2024 and 2030.

Source: hmm Ventures regulatory dataset.

The Necessity Doctrine, Chapter 4

Japan is the clearest test of whether regulated corporate procurement can drive seed-stage venture outcomes. The country’s corporate architecture, its regulatory posture, and its capital structure are coupled in a way that creates some of the strongest seed-stage barriers in developed markets. The cross-shareholding system that has historically blocked outside disruption can route capital, procurement, and validation toward regulated startups.

The paradox is that the same feature Western investors have criticised for decades may now create an advantage in regulated procurement. The driver is the AI Promotion Act 2025, the broader Economic Security Promotion architecture, and a generational handover at the corporate level that is beginning to loosen capital allocation and procurement decisions.

Japan, current state

  • AI Promotion Act: passed 28 May 2025; most provisions in force 4 June 2025; AI Strategy Headquarters chaired by the Prime Minister, established 1 September 2025.1
  • Enforcement model: no fines, no bans, no mandates. Cooperative compliance with public name-and-shame for non-compliance. METI Guidelines for Business v1.1 (March 2025); v1.2 in preparation addressing EU AI Act interoperability.2
  • Economic Security Promotion Act: in force 2023; eleven specified critical materials including semiconductors and pharmaceuticals; supply chain inspection regime active.
  • Corporate balance sheets: approximately ¥637 trillion in retained earnings on aggregate, the highest ratio of any G7 economy.3

The cross-shareholding paradox

The traditional Japanese corporate structure is built on cross-shareholding (kabushiki mochiai) and main-bank relationships. Outsiders cannot acquire significant stakes; insiders trade them at long-cycle intervals; capital reallocation is slow by design.

For thirty years, this structure was treated by Western investors as the obstacle. Activist campaigns, governance reform, the 2014 Corporate Governance Code, and the 2023 TSE prime market criteria all targeted unwinding cross-shareholdings. Some of that has happened. More of it has not.

The paradox sits in what the structure now produces. Japanese corporates retain the highest aggregate cash position in the G7. When those corporates are required by regulation to upgrade compliance, security, supply integrity, or AI governance, they deploy that capital at speed and through channels their own venture arms control.

The implication for seed-stage investors is direct. Companies that build for the regulated Japanese corporate buyer have access to a deeper procurement budget than equivalents in Europe or North America, and the procurement is filtered through corporate venture relationships rather than open RFP processes.

The AI Promotion Act and what it enables

Japan’s AI Promotion Act took an explicitly different design choice from the EU AI Act. There are no penalties, no prohibitions, no mandatory categorisations. Compliance is voluntary in form and reputational in enforcement.

That choice matters for the doctrine. The EU framework creates an enforcement perimeter; companies on the wrong side face €35 million fines. The Japanese framework creates a participation perimeter; companies inside the perimeter are more likely to receive procurement attention, regulatory cooperation, and access to the AI Strategy Headquarters coordination process. Companies outside it receive nothing punitive, but lose that advantage.

For T1 and T2 companies, the participation perimeter is the asset. METI’s AI Governance Guidelines for Business v1.1 set practical expectations: documented risk assessment, transparency to users, monitoring infrastructure, lifecycle records.2 Companies that build to these expectations are easier for regulated Japanese buyers to approve. Companies that do not, are harder.

A v1.2 update is reportedly in preparation, addressing cross-border data flows and EU AI Act interoperability. METI has not publicly confirmed scope or timing. The interoperability move, if it lands, is significant. It may reduce the amount of retooling required for a Japanese T2 vendor to sell into both Japanese and European regulated buyers.

The capital architecture

Corporate venture capital in Japan is structurally distinct from CVC elsewhere. The long-cycle horizons, the limited partner pressures of patient capital from main banks and life insurers, and the tight coupling to corporate parent procurement budgets produce a different deployment profile.

Major Japanese CVC vehicles (Sumitomo Mitsui’s CVC arm, MUFG Innovation Partners, SoftBank Ventures, NTT Docomo Ventures, JIC VGI, and others) deploy at scale into both domestic and international targets. A meaningful share of inside-Japan deployment routes through CVC vehicles to Necessities-aligned companies serving the regulated parent enterprise as customer or technology partner.4

The implication is that the Japanese seed-stage Necessities is not really a Japanese-only Necessities. It is the Necessities about access to Japanese corporate buyer relationships from any geography, with Japanese-domiciled cap table participation. Companies based in Singapore, Sydney, or Auckland can serve the Japanese T1 buyer if they enter the corporate venture relationship layer.

What the doctrine looks for

Three signals at deal level. First, an existing or imminent CVC relationship with a Japanese-listed corporate that operates in one of the Five Necessities. Second, METI guideline alignment as a procurement gate, not a marketing item. Third, a co-founder or senior commercial lead with Japanese language and procurement-process competence.

The third signal is the rarest. Most APAC-domiciled companies pursuing Japan substitute it with a partnership distribution model and a non-Japanese-speaking sales team, which produces a Japan-shaped logo on the customer slide but no procurement traction. The doctrine treats absence of the third signal as a structural risk, not a hiring gap to fix later.

Risks and counter-arguments

The Japanese demographic profile is the standard counter-argument. The population is shrinking, the workforce is aging, and the domestic addressable market is on a long-cycle decline. The doctrine accepts the demographics and bets on the corporate balance sheet rather than the consumer wallet.

Foreign investor participation in Japanese venture remains thin compared to North America or Europe. That is part of the Necessities. The entry valuation discount is real and driven by capital scarcity, and exits to North American strategic acquirers extract the valuation differential. The discount also reflects friction in cap table management, post-money rights, and exit liquidity that a foreign-domiciled fund must underwrite explicitly.

The name-and-shame enforcement model is untested. No company has yet been publicly named under the AI Promotion Act, and the political economy of doing so is uncertain. If the model proves toothless, the participation perimeter degrades into a compliance suggestion. The doctrine assumes the perimeter holds. The falsification condition is two cycles without a public naming when one is warranted.

What this market produces

A specific seed-stage profile. Companies whose Series A is co-led by a Japanese CVC and an APAC-regional fund. Whose first three logos include one Japanese listed enterprise. Whose product roadmap aligns explicitly to METI guideline categories. Whose exit path runs through an English-language strategic acquirer that values the Japanese revenue base for its durability rather than its growth rate.

That profile is rare. The candidates that match it sit at lower entry valuations than equivalent profiles in North America or Europe, because the fundraising market for them is thin. The next chapter examines Australia, where the architecture is different but the entry-valuation arithmetic produces a similar conclusion.

Chapter 5 examines Australia: where R&D incentives, superannuation capital, and the regulatory architecture of a mid-sized commodity-export economy converge.

Footnotes

  1. Government of Japan, Act on the Promotion of Research, Development and Utilization of AI-Related Technologies (AI Promotion Act), enacted 28 May 2025. Most provisions effective 4 June 2025; AI Strategy Headquarters effective 1 September 2025. ↩

  2. Ministry of Economy, Trade and Industry (METI) and Ministry of Internal Affairs and Communications (MIC), AI Guidelines for Business v1.1 (March 2025); v1.2 reportedly in preparation per industry briefings, METI confirmation pending. ↩ ↩2

  3. Ministry of Finance, Financial Statements Statistics of Corporations by Industry (法人企業統計), FY2024 annual release, September 2025. Thirteenth consecutive year of record aggregate internal reserves; excludes finance and insurance. ↩

  4. Author’s compilation from corporate venture vehicle disclosures and CB Insights Japan CVC tracker, 2025 calendar year. Specific deployment volumes and inside-Japan share require independent verification. ↩