Fujitsu Builds Sovereign AI Architecture "Takane" for Global Banking

Financial infrastructure is moving away from public cloud LLMs to insulate sensitive banking data. On July 28, Fujitsu announced it will initiate development of the "Uvance for Finance AI Transformation Platform," a dedicated proprietary AI platform for financial institutions including regional banks in Japan. The platform will begin development on August 1, 2026, with the finished product set to launch in March 2027.

The platform is built on Fujitsu Kozuchi Enterprise AI Factory, a proprietary AI platform that ensures data sovereignty and AI trustworthiness. It incorporates Fujitsu's AI technologies, including the "Takane" large language model, which is designed to handle the unique business practices, legal frameworks, and specialized terminology of financial institutions. Takane is a highly accurate Japanese LLM built with Cohere's advanced language model technology to understand and process complex technical language across specialized industries including government, finance, healthcare, and law. The platform also includes generative AI trust technologies that address security requirements such as guardrail technology for vulnerabilities, and multi-AI agents that autonomously perform tasks according to business characteristics.

The multi-agent AI systems within the network autonomously audit credit risk, verify legal documentation, and execute anti-money laundering checks in real time. A customer understanding agent and a sales activity optimization agent collaborate to support tasks such as loan approvals and document checks. The platform handles tasks including loan screening, document creation and verification, inquiry handling, branch office operations, compliance, reporting, and customer support. Guardrail technology visualizes AI reasoning and risks to meet strict security requirements. Takane will also help optimize AI usage costs through a flexible pricing model combining flat-rate and usage-based billing.

Fujitsu will support financial institutions in securely integrating AI into their operations and decision-making, transforming business processes into AI-driven workflows that consistently deliver results. The initiative reflects a broader industry shift as financial institutions move from pilot phases to full implementation, with approximately 50% of financial institutions now using generative AI in actual operations. The platform is also expected to foster a co-creation ecosystem where multiple financial institutions can accumulate and mutually utilize AI agents and operational expertise. July 2026 confirms that financial AI is moving decisively toward sovereign, specialized infrastructure—where data never leaves the institution's control, models are tailored to industry-specific requirements, and multi-agent systems handle complex workflows autonomously. The era of generic public cloud LLMs in banking is ending. The era of sovereign, industry-specific AI architecture is already here.