Data 2026-07

Sovereign AI

Announced commitments against government budgets and reported disbursements, by country.

Sovereign programme record · Japan · 2024–2030

Japan: METI ¥10T AI Strategy / RIKEN FugakuNEXT / Rapidus: $64.0B announced · 5.9% of the government-only floor disbursed

Every figure below derives from the same record the dashboard reads — this page is the programme as a citable object: the reconciliation with its denominator stated, the decomposition with its residual, and the dated evidence behind each claim.

Announced~$64.0B
Government-only floor$7.90B
Committed~$7.90B
Disbursed~$470M

The reconciliation. Four figures on one dollar scale, each bar a share of the largest. The reality ratio above divides disbursed by the government-only floor ($7.90B), and the field it divides by is stated so the division is reproducible. A hatched track is an absent figure carried with its reason — a government that does not publish disbursement is a finding, not a zero. Tildes mark estimated figures.

Pipeline stage: Appropriated (high confidence) — 6% of announced disbursed. The stage is a Scrutica derivation (tier 4), computed from the figures above, never asserted by the source.

What the announced figure is made of

Component	Kind	Amount
METI ¥10T public support through fiscal 2030	Government	$64.0B
FY2026 budget: ¥1.23T ($7.9B) for semiconductors and domestic AI	Government	$7.90B
METI direct: $740M to 6 AI compute firms	Government	$740M
Private sector (SoftBank JPY 150B ~$960M, NTT, etc.)	Private	—

The residualThe amount-bearing components sum to $64.0B against the $64.0B headline — they reconcile.

Hardware on record

Model	Count	Status	Owner
NVIDIA Blackwell	2,140	procurement_active (spring 2026 target)	RIKEN (government)

Empanelled partners · 17

Partner	GPUs	Model	Stage	Award
SAKURA Internet	10,800	NVIDIA H100 + NVIDIA HGX B200	awarded	Primary source · 2024-04-19
KDDI Corporation	—	NVIDIA HGX + planned NVIDIA GB200 NVL72	awarded	Primary source · 2024-04-19
Highreso	1,600	NVIDIA H200	awarded	Primary source · 2024-04-19
RUTILEA	1,000	NVIDIA Hopper (H100/H200)	awarded	Primary source · 2024-04-19
GMO Internet Group	—	NVIDIA H200	awarded	Primary source · 2024-04-19
Sakana AI	—	—	empanelled	Primary source · 2024-02-02
Turing Inc.	—	—	empanelled	Primary source · 2025-07-15
Preferred Elements, Inc.	—	—	empanelled	Primary source · 2024-02-02
Kotoba Technologies Japan, Inc.	—	—	empanelled	Primary source · 2024-02-02
SoftBank Corp.	—	—	empanelled	Primary source · 2024-02-02
ABEJA, Inc.	—	—	empanelled	Primary source · 2024-10-10
Matsuo-Iwasawa Laboratory, The University of Tokyo	—	—	empanelled	Primary source · 2024-02-02
Stockmark Inc.	—	—	empanelled	Primary source · 2024-02-02
Ubitus K.K.	256	NVIDIA H100	empanelled	Primary source · 2024-10-10
AI inside Inc.	—	NVIDIA H100 (via AWS EC2 P5)	empanelled	Primary source · 2024-10-10
Rakuten Group, Inc.	—	—	empanelled	Primary source · 2025-07-15
Alivexis, Inc.	—	—	empanelled	Primary source · 2025-07-15

External dependence

NVIDIA hardware: high · TSMC fabrication: high · US technology: high · BIS jurisdiction reach: low. Dependence levels are Scrutica classifications (tier 4) from the programme’s own disclosures — which hardware it buys, where it fabricates, and whose export jurisdiction its supply chain crosses.

Named partners: NVIDIA, Fujitsu, Sakura Internet, KDDI, SoftBank, GMO, Highreso, RUTILEA, RIKEN, MEXT, Rapidus.

METI committed ¥10 trillion ($64B) public support through fiscal 2030. Target: 60 EFLOPS by fiscal 2027. FY2026 budget: ¥1.23 trillion ($7.9B) for semiconductors and domestic AI. Rapidus 2nm GAA milestone July 2025. RIKEN 2,140 Blackwell GPUs targeting spring 2026. FugakuNEXT: 600+ EFLOPS FP8 target ~2030. METI: $740M to 6 AI compute firms. SoftBank: JPY 150B (~$960M) AI infrastructure.

What this record flags about itself · 7

METI committed ¥10 trillion ($64B) public support through fiscal 2030
FY2026 budget: ¥1.23 trillion ($7.9B) for semiconductors and domestic AI
Target: 60 EFLOPS by fiscal 2027
Rapidus 2nm GAA milestone July 2025 — on track for pilot production
FugakuNEXT 600+ EFLOPS is 2030 design target, not current capacity
METI $740M to 6 firms per NVIDIA blog (Tier 1); original JPY 72.5B ($470M) was initial tranche to 5 firms
SoftBank JPY 150B (~$960M) AI infrastructure investment per NVIDIA blog (Tier 1)

The evidence trail

7 cited sources

METI ¥10T strategy and Rapidus 2nm GAA milestone — Research database · no accessed date on record
NVIDIA Japan sovereign AI partnership — Primary source · no accessed date on record
RIKEN 2,140 Blackwell GPUs, spring 2026 target — Primary source · accessed 2026-03-28
RIKEN official: FugakuNEXT with Fujitsu + NVIDIA, 600+ EFLOPS target, ~2030 — Primary source · accessed 2026-03-28
METI ¥72.5B, Sakura Internet ¥50.1B, KDDI ¥10.2B — Research database · accessed 2026-03-28
NVIDIA blog: FugakuNEXT details, MEXT backing — Primary source · accessed 2026-03-28
METI program cloud providers NVIDIA infrastructure — Primary source · accessed 2026-03-28

Machine surfacesThis record is one object under one id: the evidence-trail feed and the derivation-chain export serialize the same programme this page renders.