Fujitsu launches multi‑AI service to modernize legacy systems 40% faster

Fujitsu’s new service is a reminder that the bottleneck for AI adoption in large enterprises isn’t just model quality — it’s the mountain of brittle legacy systems underneath. By productizing a modernization pipeline built around multiple frontier models plus Fujitsu’s own Takane LLM, the company is effectively wrapping “AI as refactoring engine” into a service line that speaks CIO language: shorter timelines, lower risk and less dependence on scarce COBOL or mainframe specialists.

Strategically, this moves generative AI closer to the core transaction systems of banks, insurers and public agencies, which have historically been walled off from LLMs by compliance and stability concerns. If Fujitsu can reliably automate large portions of code analysis and conversion, it becomes a powerful distribution vector for Anthropic and OpenAI models into high‑value, highly regulated environments. It also strengthens Japan’s domestic AI stack by pairing local models with global ones under a unified orchestration layer.

For the race to AGI, the significance is indirect but real: the more legacy code gets translated into AI‑navigable form, the easier it becomes to build agents that understand enterprise systems end‑to‑end, not just sit on top as chatbots.