NEC partners with Anthropic on Claude AI while pursuing "four sovereignties"

NEC has announced a global partnership with Anthropic, maker of Claude, an AI language model that understands and generates text. Under the partnership, NEC will integrate Claude into its enterprise solutions and will have access to Claude Mythos, a large language model that NEC can use as the foundation for building AI applications customized to its customers' needs. The partnership centers on what NEC calls "AI nativity"—the ability to embed Claude into NEC's own systems and business processes.

The announcement comes at a time when global AI leaders have already made significant decisions about how to deploy their most advanced models. Anthropic's Claude Opus, Google's Gemini, and OpenAI's models are all undergoing security reviews and being made available across a broad customer base. But the security scrutiny is complex and varied, and Anthropic's Mythos model—which is highly capable—has not yet been subject to the same depth of external security testing as earlier models. Broadly available security standards have not yet been established, leaving organizations to design their own risk frameworks.

NEC President and CEO Tsutomu Suzuki has been emphatic about the terms under which the partnership was struck. In an interview, Suzuki articulated four areas where NEC refuses to compromise: data sovereignty (control over where enterprise data is stored and used), technology sovereignty (the ability to customize and understand the AI model), business sovereignty (independence in deciding how and where to deploy AI), and human sovereignty (maintaining human oversight and transparency in AI decision-making). Suzuki notes that while Anthropic, as a global company, naturally prefers to negotiate standardized partnership terms that apply across all its major customers, NEC was able to complete the negotiation in just three months by insisting on tailored governance from the start.

Suzuki's position reflects a broader concern in the Japanese enterprise market. While many companies are eager to adopt cutting-edge AI, they are wary of ceding control over critical business systems to foreign vendors. Suzuki observes that even as technology giants move at high speed to deploy standardized AI, the underlying security risks are still unresolved—a fact that makes it essential for enterprises to retain the ability to customize the model, validate its behavior in their own systems, and maintain clear visibility into how it makes decisions. He points out that this is not a problem that can be solved by technology and software alone; it requires organizational discipline, internal expertise, and a clear understanding of which decisions in the enterprise architecture can be delegated to AI and which must remain under human control.

Suzuki is particularly focused on ensuring that NEC can access Mythos and adapt it to Japan-specific regulatory and security requirements. He acknowledges that while some critical infrastructure has been opened to foreign technology and security reviews, the governance of the most advanced AI models—especially those used in sensitive business and government applications—remains an open question in Japan. For this reason, Suzuki believes it is essential that NEC understand Anthropic's technology roadmap and be able to influence how the partnership evolves over time. He also emphasizes the importance of maintaining a dialogue with Anthropic about the global speed of AI deployment and the need to align that pace with the governance and security standards that Japanese enterprises and regulators are still developing.

NEC's partnership with Anthropic represents a deliberate attempt to balance access to world-class AI technology with the preservation of Japanese enterprise control and regulatory compliance. While Anthropic—like Google with Gemini and OpenAI with their models—has built Claude to be accessible across a wide range of customers, Suzuki is negotiating for something different: the ability to adapt the model to NEC's own systems and security standards rather than adopting a one-size-fits-all offering. This points to a deeper tension in the AI industry. Global AI leaders have optimized for speed and scale by standardizing their model deployments, but Japanese enterprises and regulators increasingly require the ability to understand, customize, and govern the AI systems embedded in their critical business processes.

Suzuki's emphasis on "four sovereignties" is not merely rhetorical. It reflects real business risks: data breaches if customer information flows through global cloud infrastructure without clear controls, vendor lock-in if the AI model cannot be customized to Japanese regulatory requirements, and loss of strategic autonomy if business-critical AI decisions are opaque or centrally controlled by a U.S. company. By completing the negotiation in three months despite Anthropic's standard terms, Suzuki has signaled both urgency (NEC cannot afford to fall behind in AI deployment) and leverage (Anthropic sees value in the Japanese enterprise market).

The outcome suggests that large Japanese companies may be able to negotiate middle-ground partnerships with global AI firms—adopting their technology while carving out space for domestic governance, security review, and customization. However, success will depend on whether NEC can actually build the internal expertise and security infrastructure to manage that boundary without creating bottlenecks or costs that offset the partnership's benefits.