NextFin News - On December 24, 2025, Japan’s Fujitsu Limited announced the launch of Fujitsu Kozuchi Physical AI 1.0, a pioneering platform co-developed with U.S. semiconductor heavyweight Nvidia. This platform integrates multiple AI agents to automate confidential and complex business workflows with enhanced security. The development marks the first achievement of their partnership, initiated in October 2025, and is headquartered in Kawasaki, Japan.
The platform features a multi-AI agent framework built on Fujitsu’s proprietary Takane large language model (LLM) and Nvidia’s NIM microservices architecture. It offers a visual user interface for designing complex workflow automation applicable across multiple departments in enterprises. With focus areas including procurement compliance and regulatory analysis, specialized AI agents reduce manual workload by approximately 50%, accelerating processes that involve sensitive document interpretation and multi-party supplier interactions.
Fujitsu’s integration ensures secure confidentiality via its Secure Inter-Agent Gateway technology, which enables safe collaboration among AI agents developed by various vendors, crucial for protecting enterprise-private data in inter-company workflows. Trial deployments within Fujitsu’s purchasing division have demonstrated significant efficiency gains, signaling potential for broader industry adoption by the end of fiscal 2025.
Under U.S. President Trump’s administration, amid heightened global AI competition and growing demand for technology sovereignty, this Japan-U.S. collaboration addresses both innovation acceleration and strategic autonomy. Fujitsu aims to expand the platform to physical AI domains, enabling AI agents to interact directly with real-world environments through robotics.
The development arrives at a critical moment as agentic AI, despite advances, faces adoption barriers rooted in trust, security, and industry-specific customization. Fujitsu and Nvidia’s platform confronts these challenges by allowing tailored AI agents that securely manage sensitive information and maintain workflow robustness—considerations paramount for industries such as pharmaceuticals, manufacturing, and finance.
Analyzing the drivers behind this initiative, it reflects Fujitsu’s strategic pivot to embed AI deeply in its Uvance Dynamic Supply Chain offerings, leveraging multi-agent collaboration to optimize cross-company data flows. The partnership synchronizes Nvidia’s expertise in AI infrastructure acceleration with Fujitsu’s nuanced understanding of domain-specific AI applications, creating a technically superior and commercially viable platform.
Financially, Fujitsu's forecasts anticipate revenue growth to ¥3,829 billion and earnings of ¥339.8 billion by 2028, which the new AI platform seeks to underpin by enhancing contract scalability and recurring service margins. Empirical results from internal trials, such as a 50% reduction in order confirmation delays and 50% faster AI inference speed via Nvidia NIM compatibility, substantiate efficiency gains.
Looking forward, the platform’s expansion into physical AI, facilitating autonomous real-world task execution through robotics, aligns with broader industry trends towards AI-driven automation spanning digital and physical layers. This evolution could disrupt traditional labor-intensive sectors by creating an ecosystem where multiple AI agents coordinate seamlessly across virtual and tangible domains.
This announcement also underscores geopolitical and economic dimensions. As global supply chains face fragmentation risks and increasing regulatory complexities under U.S. President Trump's policies, solutions offering sovereignty, security, and resilience become critical. Fujitsu and Nvidia’s collaboration strengthens regional technological autonomy, offering an alternative to Western- or Chinese-dominated AI ecosystems.
Moreover, by supporting highly confidential workflows through federated agentic AI architecture, the platform anticipates future enterprise demands for privacy-preserving AI, fulfilling strict compliance mandates and data governance. This is essential as increasing legal scrutiny and industrial standards shape AI adoption worldwide.
Strategically, this initiative advances the transition from isolated AI models to interoperable, multi-agent ecosystems. This represents a new AI paradigm shift, characterized by agent collaboration, domain specialization, and continuous autonomous learning, which Fujitsu plans to realize fully by the end of fiscal 2025. Should this vision mature, it will set a benchmark for AI infrastructure providers globally.
In conclusion, the Fujitsu-Nvidia multi-agent AI platform marks a significant milestone in enterprise AI automation. By combining secure multi-agent frameworks with performance-optimized hardware support, the platform addresses longstanding challenges in AI adoption for confidential, complex workflows. Its ongoing development and planned expansion into physical AI domains suggest transformative impacts on industrial automation, operational resilience, and cross-industry collaboration into the coming years under the U.S. President Trump administration's focus on technological leadership.
Explore more exclusive insights at nextfin.ai.
