METI's FY2026 budget allocates ¥1.23 trillion ($7.9B) for AI and semiconductors — nearly quadrupling previous support. ¥387.3B targets domestic foundation models, data centers, and physical AI for robotics. Separately, ¥1 trillion ($6.3B) sovereign AI initiative funds a new entity formed by ~10 companies including SoftBank to build Japan's most advanced foundation model with 1 trillion parameters.
Fujitsu began manufacturing NVIDIA-based AI servers domestically under Made in Japan sovereign AI initiative. Targets government and enterprise customers requiring data sovereignty and supply chain security.
Japan's Growth Strategy Headquarters, chaired by PM Takaichi, identifies 61 specific products and technologies across 17 strategic sectors for priority investment. The list includes 27 technologies under early-stage review: physical AI systems, regenerative medicine, quantum computing, and marine drones. Part of ¥21.3T stimulus package with ¥7.2T earmarked for strategic investment in AI, semiconductors, quantum, defense, and energy. Semiconductor sales target: fivefold increase to ¥40T by 2040.
Microsoft commits $10B to Japan over 2026-2029 — its largest-ever Japanese investment — explicitly framed as sovereign AI infrastructure. Partners with Sakura Internet (which receives ¥50.1B / $324M from METI for AI supercomputer build) and SoftBank for in-country GPU supply. Sakana AI closes Series B at ¥32B ($200M) on April 9 to accelerate sovereign AI mission. METI joint venture with 10+ domestic firms (SoftBank, Preferred Networks) gets $6.6B over 5 years from FY2026.
SoftBank committed $12.7B of its own capital to build massive AI data centers across Hokkaido and Osaka, emerging as Japan's primary sovereign compute provider. This supplements the METI $6.6B government allocation and Microsoft's $10B investment, bringing total committed sovereign AI infrastructure investment in Japan to over $30B.
SoftBank, NEC, Honda, and Sony jointly established Nihon AI Foundation Model Development Co. on April 12, targeting a trillion-parameter physical AI foundation model trained on Japanese data for real-world control tasks (robotics, automotive, manufacturing). Nippon Steel, Kobe Steel, Mitsubishi UFJ, SMBC, Mizuho are minority investors. METI/NEDO providing ¥1 trillion over 5 years.
SoftBank, NEC, Honda, and Sony jointly formed Japan AI Foundation Model Development company to build a trillion-parameter Physical AI model (robots, autonomous vehicles, industrial machines — not chatbots). Investors include Nippon Steel, Kobe Steel, MUFG, SMBC, Mizuho. NEDO earmarking ~¥1T ($6.28B) in AI support over 5 years from FY2026. SoftBank committing additional $12.7B for AI data centers in Hokkaido and Osaka.
SoftBank, NEC, Honda, Sony and ~10 partners (incl. Nippon Steel, MUFG, Mizuho) formed Japan AI Foundation Model Development Co. to build a trillion-parameter physical AI model for robotics/manufacturing. Data stays in Japan — no foreign cloud. NEDO has earmarked ¥1T over 5 years.
SoftBank announced on May 25, 2026 it will launch an 'AI Data Center GPU Cloud' powered by its in-house 'Infrinia AI Cloud OS' as part of its neocloud business, with a beta live from announcement day and general availability in October 2026. Built on NVIDIA GB200 NVL72 hardware, it offers Kubernetes-as-a-Service (KaaS) for multi-tenant environments and Inference-as-a-Service (Inf-aaS) for LLM APIs, positioned as sovereign GPU infrastructure for Japan's neocloud market. Initial rollout is internal-first across SoftBank group companies before commercial release.
SoftBank, Oracle, and SB Intuitions began the sequential rollout of generative AI services powered by Sarashina — SB Intuitions' homegrown Japanese-language LLM — on Cloud PF Type A, SoftBank's Oracle Alloy-based sovereign cloud platform. Sarashina is positioned for high Japanese-language processing and Japan-specific cultural/customary context. The rollout pairs with the May 25 launch of SoftBank's Infrinia AI Cloud OS GPU Cloud (beta now, GA October 2026).