AI as Infrastructure

The organizations making the most meaningful progress with AI are increasingly treating it as core infrastructure instead of a collection of disconnected point solutions layered onto existing workflows. Without an infrastructure-first mindset, organizations risk accumulating short-lived tools that cannot scale across the enterprise, rapidly increasing costs, and creating activity and buzz around AI without developing a durable AI strategy that drives positive ROI.

As AI moves beyond copilots and chatbots into agentic systems capable of reasoning, acting, coordinating workflows, and operating across complex environments, this infrastructure approach becomes particularly important for governments and public-sector organizations.

Japan Has a Different Set of Requirements

Many of the companies developing today's frontier AI models and services are based in the United States, while governments and enterprises elsewhere operate under their own requirements around data privacy, sovereignty, security, regulation, procurement, and deployment.

For Japanese public-sector organizations in particular, relying exclusively on AI services delivered through externally hosted cloud environments and APIs may not satisfy requirements around where data resides, how systems are governed, or who ultimately controls the underlying technology.

Japan therefore needs an AI approach that provides access to rapidly advancing capabilities while accommodating the security, localization, governance, infrastructure, and long-term control its institutions require. Japan’s Digital Agency is already moving in this direction through its Government AI initiative (“GENAI”), which combines government-wide AI infrastructure, in-house development, domestic model support, government data, and security requirements. https://www.digital.go.jp/en/policies/genai

The Frontier-Lab Model Doesn’t Solve Everything

The frontier labs have driven extraordinary advances in AI capability, but relying exclusively on proprietary models delivered through centralized infrastructure can create strategic dependencies around access, cost, policy, and technology roadmaps.

For governments and other society-critical organizations, the question is not whether to use frontier technology, but how to benefit from it without becoming permanently dependent on a single model provider, cloud platform, or foreign technology ecosystem. The alternative is an architecture that preserves optionality, including the ability to choose models, control infrastructure and data, govern system behavior, and adapt as both the technology and the geopolitical environment change.

For Lazarus, solving that challenge means working alongside the institutions and technology providers already serving its most critical environments.

Why Japan’s Technology Leaders Are Working With Lazarus

Lazarus has spent years working alongside some of Japan’s largest and most established technology organizations, including NTT, BIPROGY, and OKI, whose reach spans municipal government, critical infrastructure, regulated industries, and defense.

These partners bring deep institutional knowledge, trusted customer relationships, and decades of experience operating inside Japan, while Lazarus brings expertise in building and deploying agentic AI systems with independence and control built into the architecture. The common objective is to give Japanese organizations access to advanced AI capabilities without forcing them to sacrifice the governance, sovereignty, and operational control that their most important work demands.

OKI as an Example

Earlier this year, Lazarus announced a strategic partnership with OKI, an organization with more than a century of experience building and supporting communications, infrastructure, and technology systems in Japan. The partnership brings together OKI’s deep experience inside critical Japanese environments with Lazarus’s expertise in building a new generation of agentic AI systems.

We believe that combination can accelerate the responsible adoption of advanced AI across mission-critical areas spanning social infrastructure, the public sector, and defense, and it reflects the same complementary model we pursue with partners across the broader public sector as well as regulated industries.

What “Independence” Actually Means

AI independence does not mean rejecting frontier models or isolating an organization from the broader AI ecosystem; it means retaining meaningful control over how those technologies are selected, deployed, governed, and replaced. In practice, that means model choice, infrastructure choice, control over data, flexible deployment, transparent testing and governance, and the ability to determine how an agentic system behaves within an organization’s environment.

Flexibility is especially important internationally, where organizations cannot assume that today's providers, access policies, pricing structures, regulatory environment, or geopolitical relationships will remain unchanged.

What Lazarus offers, in conjunction with our partners, are fully independent and sovereign AI capabilities. From the foundational intelligence layer (models, hardware, serving, kernels) up to execution with agentic orchestration, autonomous solutions, and governance, we designed our entire technology stack to give organizations lasting control over how, where, and on whose terms their AI operates.  And this is all built on top of a leading AI research team widely cited for their work advancing model independence and scalability (Economist article citing Lazarus AI’s Chief Scientist Eric Hartford).  

Why This Matters in Japan

AI sovereignty and technological dependency are strategic questions for governments worldwide, and they take on particular importance in countries seeking greater domestic control over frontier AI models and infrastructure.

For Japan, maintaining greater control over how AI is deployed across government, critical infrastructure, defense-adjacent environments, and regulated industries can support resilience while letting institutions adopt new capabilities more rapidly.

The goal is not technological isolation; it is ensuring that increasingly important national capabilities do not depend entirely on decisions, infrastructure, or policies controlled somewhere else.

Why the Timing Matters Now

The rate of improvement in AI means the cost of waiting is changing. Capabilities that were experimental a year ago are increasingly becoming practical systems for software development, cybersecurity, research, operations, and knowledge work. 

Organizations that build flexible AI infrastructure now can adopt those advances as they emerge, while organizations locked into rigid applications or lengthy deployment cycles risk falling behind competitors and peers that can move faster.

The challenge is therefore not simply to select the best model available today, but to build an operating architecture that allows organizations to continuously incorporate better models and agentic capabilities as the frontier advances.

What's Next

Our work in Japan is strongest when Lazarus's expertise in agentic AI is paired with the experience, reach, infrastructure, and trust of organizations already deeply embedded in Japanese society.

Together, our role is to bring advanced AI into consequential environments while preserving control over the models, infrastructure, data, governance, and systems those organizations and their customers depend on.

Japan now has an opportunity to participate fully in the next generation of AI. Not simply by consuming technology developed elsewhere, but by deploying, governing, adapting, and advancing it on its own terms.

Build the Future of AI on Your Terms

The next generation of AI will be built by organizations willing to rethink how intelligence is deployed, governed, and put to work in the world’s most consequential environments.

Lazarus works alongside technology providers, infrastructure leaders, integrators, and other strategic partners to turn that ambition into practical, production-ready AI systems built for independence, flexibility, and long-term control.

If you’re helping shape what comes next in AI, we’d like to build it with you.

‍