Skip to content

How NTT, Fujitsu, and NEC Signal Different Enterprise AI Strategies

For / Key Points

For: Enterprise AI leaders, engineering managers, business planners, and CoE owners evaluating generative AI, AI agents, or AI coding tools in large organizations.

Key Points:

  • NTT, Fujitsu, and NEC all talk about enterprise AI, but their strategic centers of gravity differ
  • NTT points toward domestic LLM infrastructure, Fujitsu toward business AI components, and NEC toward customer DX/AX tied to engineering transformation
  • Official announcements should be read as positioning signals, not as proof that outcomes have already been achieved

Read Official Announcements as Strategy Signals, Not Outcome Proof

NEC's 30,000-person Claude Code rollout is a weak signal if read in isolation. It can collapse into a simple headline: a major company is giving many employees access to an AI coding tool.

The picture changes when NTT, Fujitsu, and NEC are placed next to each other. All three companies are talking about AI adoption, but they are not emphasizing the same layer. NTT is foregrounding domestic LLM infrastructure and safe social implementation. Fujitsu is presenting AI as components that can be embedded into business workflows. NEC is connecting customer DX/AX with internal engineering transformation.

The factual base here is limited to official PR and IR-adjacent materials. Those materials are still company announcements, not independent proof of business impact. The comparison here is not about who has already won. It is about which layer of enterprise AI each company appears to be betting on.

Where the Three Strategies Differ

CompanySignal from official announcementsMain materialsWhat reader companies should inspect
NTTSocial infrastructure, domestic LLMs, low-cost operation, AI governancetsuzumi 2, closed and on-premise operation, Chief AI Officer, AI CharterHow much model sovereignty, data sovereignty, and operating cost control matter
FujitsuBusiness AI platforms, domain-specific models, auditability and trustFujitsu Kozuchi, Takane, enterprise generative AI framework, Kozuchi AI AgentWhether AI can be designed as workflow components rather than isolated tools
NECCustomer DX/AX, Client Zero, AI Platform Service, engineering transformationBluStellar, cotomi, AI Platform Service, Anthropic and Claude Code collaborationWhether internal use can feed back into customer offerings

The distinction is already visible from this table. NTT asks how AI can run on a trusted foundation. Fujitsu asks how AI can be embedded into business operations. NEC asks how AI can connect customer transformation with internal transformation.

NTT: AI as Social Infrastructure and Domestic Foundation

NTT's AI positioning is less about a single productivity tool and more about foundation-layer control. Its October 2025 announcement for tsuzumi 2 frames the model around Japanese-language performance, electricity consumption, operating cost, and security risks when handling confidential information1.

tsuzumi 2 is presented as a lightweight, purely domestic model that can run inference on a single GPU and operate at low cost in on-premise or private-cloud environments1. NTT's AI page also refers to AI risk management, a Chief AI Officer, and the NTT Group AI Charter2.

The strategic signal is clear: NTT is emphasizing the foundation layer of model, operating environment, and governance. That positioning is naturally relevant to public-sector, healthcare, finance, local-government, and closed-network environments where data location and accountability matter.

For reader companies, the question comes before tool selection. Where will sensitive data be processed? What cost structure can be sustained? How much control over the model and operating environment is required? The heavier those questions are, the more NTT's framing becomes relevant.

Fujitsu: AI as Components Embedded Into Business Workflows

Fujitsu's official materials present AI as a set of components that can be embedded into business operations. Fujitsu Kozuchi is described as a collection of AI service areas, including generative AI, AutoML, predictive analytics, vision, text, AI Trust, and XAI3.

In June 2024, Fujitsu announced an enterprise generative AI framework with knowledge-graph extended RAG, technology for combining multiple models, and generative AI auditing for compliance with laws and corporate rules4. In September 2024, Takane was announced as a Japanese-language LLM for secure enterprise use, integrated with Kozuchi and Data Intelligence PaaS5. In October 2024, Fujitsu Kozuchi AI Agent was announced as an AI service that can work autonomously and in collaboration with humans on high-level tasks6.

Taken together, Fujitsu is not simply saying that enterprises should adopt a large general-purpose AI. It is presenting a path where business data, business rules, auditability, and AI agents are combined inside workflows. That makes it easier to discuss AI value at the level of contract review, support desks, negotiations, manufacturing, legal work, and similar business processes.

For reader companies, the issue is whether AI adoption can move from chat-tool deployment to workflow design. If business rules, audit trails, exception handling, and model selection are not designed together, this direction can easily become a platform purchase without operational change.

NEC: AI Connected to Customer DX/AX and Engineering Transformation

NEC's positioning is slightly different from NTT's and Fujitsu's. NEC also has foundation and service elements: cotomi, BluStellar, and AI Platform Service. cotomi is positioned as a key technology for BluStellar, with announcements around specialized business use and improved GPU efficiency7. NEC's April 2026 AI Platform Service announcement describes more than 100 service functions, including AI agents, models, connection protocols, and data integration, intended to accelerate AX, or AI Transformation8.

The difference is that NEC is connecting those service layers with internal engineering change. Its Anthropic collaboration places industry-specific AI for finance, manufacturing, and local government beside Claude and Claude Code use in BluStellar Scenario, Claude availability for roughly 30,000 NEC Group employees, Client Zero, and an AI-native engineering team using Claude Code910. Here, Client Zero means using NEC's own organization as the first customer.

In other words, NEC is not only presenting AI as a customer-facing DX/AX product set. It is also presenting AI as something that changes its own development organization and then feeds back into customer offerings. That is a different vector from NTT's foundation emphasis and Fujitsu's workflow-component emphasis.

This is interesting, but the outcome cannot be judged from the announcement. The scale of 30,000 people is visible, but the real value should be measured by review load, defect rate, development lead time, incident rate, and reusable assets that flow into customer delivery.

NEC Looks Like a Different Bet, Not a Simple Outlier

When the three companies are compared, NTT and Fujitsu are easier to place. NTT is about trusted domestic AI infrastructure. Fujitsu is about platformizing business AI.

NEC adds Client Zero and engineering transformation to the mix. That means the Anthropic and Claude Code collaboration is less compelling if read only as a developer-tool rollout. It makes more sense as one component that connects BluStellar, AI Platform Service, industry-specific AI, and internal engineering transformation.

This does not mean NEC is clearly ahead or clearly at risk. It means the bet is different.

One path is to own the foundation. Another is to assemble business AI components. A third is to connect internal transformation with customer-facing AI delivery. Which path is right depends on the company's starting point.

What Reader Companies Should Decide Before Selecting Tools

The takeaway for companies evaluating enterprise AI is straightforward. Before selecting a tool, decide which layer of AI adoption matters most.

Layer to decide firstCore questionSimilar framing
Foundation and sovereigntyHow much control is needed over models, data, closed environments, cost, and governance?NTT-like
Business implementationWhich workflows will AI be embedded into: contracts, support, manufacturing, legal, sales?Fujitsu-like
Organizational transformationHow will development, reviews, CoE, Client Zero, and customer delivery be connected?NEC-like

Skipping this step makes tool decisions vague. An organization may adopt Claude Code, Copilot, or a domestic LLM and then evaluate success by whether people liked it, used it, or talked about it.

If the layer is decided first, the metrics become clearer. For foundation, inspect data boundaries, operating cost, and auditability. For business implementation, inspect cycle time, exception handling, quality, and explainability. For organizational transformation, inspect review load, defect rate, reusable assets, and customer-delivery feedback loops.

Summary

NTT, Fujitsu, and NEC show that Japanese enterprise AI strategy is not one thing. NTT appears focused on social infrastructure and domestic AI foundations. Fujitsu appears focused on business AI platforms. NEC appears focused on customer DX/AX connected to engineering transformation.

NEC's 30,000-person Claude Code rollout is therefore better read as part of a Client Zero and customer-AI delivery strategy than as a simple AI coding-tool rollout. At the same time, official announcements do not prove impact. The relevant question is what improves in actual operations, not how large the announcement sounds.

The practical question is not "which AI is strongest?" It is whether your organization wants AI at the foundation layer, inside business workflows, or as part of organizational transformation.

  1. NTT, tsuzumi 2 announcement (2025-10-20). https://group.ntt/en/newsrelease/2025/10/20/251020a.html ↩↩

  2. NTT, "About AI at NTT." https://group.ntt/en/group/ai/ ↩

  3. Fujitsu, "Fujitsu Kozuchi." https://www.fujitsu.com/global/services/kozuchi/ ↩

  4. Fujitsu, enterprise generative AI framework announcement (2024-06-04). https://www.fujitsu.com/global/about/resources/news/press-releases/2024/0604-01.html ↩

  5. Fujitsu, Takane announcement (2024-09-30). https://www.fujitsu.com/global/about/resources/news/press-releases/2024/0930-01.html ↩

  6. Fujitsu, Fujitsu Kozuchi AI Agent announcement (2024-10-23). https://www.fujitsu.com/global/about/resources/news/press-releases/2024/1023-01.html ↩

  7. NEC, NEC cotomi announcement (2024-11-27). https://www.nec.com/en/press/202411/global_20241127_02.html ↩

  8. NEC, AI Platform Service announcement (2026-04-24). https://jpn.nec.com/press/202604/20260424_01.html ↩

  9. NEC, Anthropic collaboration announcement (2026-04-23). https://www.nec.com/en/press/202604/global_20260423_01.html ↩

  10. Anthropic, NEC collaboration announcement (2026-04-24). https://www.anthropic.com/news/anthropic-nec ↩