知能化されたシステム経済 … Japan’s next wave of innovators … Abeja to Ajinomoto, CarbonX and LayerX, ExaWizards and beyond … the quiet construction of a systems intelligence economy

May 19, 2026

There is a tendency, when talking about innovation, to search for noise.

We look for the visible signals: breakout consumer apps, billion-dollar valuations, charismatic founders on global stages, or sudden technological leaps that appear to rewrite the rules overnight. For years, that narrative has been dominated by Silicon Valley and, more recently, China’s platform giants.

Japan has never fitted into that mould. And yet something more profound is now underway in the vibrant cities of Tokyo, Osaka, Yokohama and beyond. Less visible, but arguably more structurally significant.

I have always been fascinated by Japan for the way it blends calmness with relentless determination. There is a quiet discipline in its culture—an obsessive yet philosophical mindset that runs from Zen monks to marathon runners, from meditative cherry blossom landscapes to the precision of the bullet train. It is a society where refinement is a lifelong pursuit, where craft becomes identity.

I have read and reread Adharanand Finn’s The Way of the Runner, which captures the Japanese psyche so well.

This is a mindset that has shaped generations of innovators, from Sakichi Toyoda’s mechanical ingenuity to Akio Morita’s global imagination. It lives on in brands like Kikkoman and Hello Kitty—simple on the surface, yet deeply intelligent in design, meaning, and enduring cultural resonance.

Today, Japan is not experiencing an innovation “boom” in the conventional sense. It is undergoing an innovation reconfiguration. The centre of gravity is shifting away from consumer platforms and toward what might best be described as systems intelligence: the embedding of artificial intelligence, robotics, materials science, and data systems into the physical and organisational infrastructure of the economy itself.

This is not disruption as spectacle. It is reconstruction as design discipline. And at the centre of this transformation are a small number of companies that, taken together, begin to define a new Japanese innovation model.

Mujin

Origin

Founded in Tokyo in 2011
Emerged from robotics and control engineering research
Built by engineers focused on industrial automation challenges

Activity

Develops AI software for industrial robots
Operates in logistics, warehousing, and manufacturing environments
Deploys globally across automation-heavy industries

Innovation

Creates “physical AI” that replaces manual robot programming
Enables real-time autonomous decision-making in machines
Builds a universal intelligence layer for industrial robotics

To understand where Japan is heading, it is useful to begin not with software, but with machines.

Mujin represents one of the clearest expressions of Japan’s new industrial logic. The company works in a domain that is deceptively simple in appearance—robotics for factories and warehouses—but its ambition is far more foundational.

For decades, industrial robots have been powerful but constrained. They execute tasks with precision, but only within tightly pre-programmed environments. Any variation—an object slightly out of place, a change in lighting, a different product shape—requires reprogramming.

Mujin’s breakthrough is to remove that constraint altogether. Its systems allow robots to perceive their environment and make decisions dynamically, in real time, without explicit human programming for each task. In effect, it transforms industrial robots from scripted machines into adaptive agents.

But what makes Mujin particularly important is not just what it builds, but how it frames the problem. It does not see itself as a robotics company in the traditional sense. It sees itself as an infrastructure company for physical intelligence. That distinction matters. It implies permanence, scalability, and systemic relevance. Mujin is not trying to build better robots. It is trying to define the intelligence layer through which global automation will operate.

There is a deeply Japanese logic here: an obsession with robustness, a preference for systems that fail gracefully—or ideally, do not fail at all—and a long-term orientation toward industrial reliability rather than rapid iteration. It is innovation expressed as engineering discipline rather than creative rupture.

Sakana AI

Origin

Founded in Tokyo by former global AI researchers
Emerged from deep learning and systems research backgrounds
Built as an alternative to large-scale model orthodoxy

Activity

Develops AI systems inspired by nature and evolution
Focuses on adaptive, efficient intelligence architectures
Works on next-generation foundation model alternatives

Innovation

Rejects pure scaling in favour of evolutionary intelligence
Explores collective and distributed AI systems
Reframes intelligence as emergent rather than centralised

If Mujin represents intelligence embedded in machines, Sakana AI represents a more radical shift: intelligence as a living system.

Sakana AI emerged from researchers who were deeply embedded in the global frontier of artificial intelligence, yet chose to step away from the dominant assumption that progress in AI is primarily a function of scale—more data, more parameters, more compute. Instead, Sakana AI asks a different question: what if intelligence is not something that is simply scaled up, but something that evolves?

Its approach draws inspiration from natural systems—collective behaviour in fish schools, evolutionary adaptation, distributed decision-making. In this framing, intelligence is not a monolithic structure but an emergent property of many interacting parts.

This is not merely a technical deviation. It is a philosophical one. Where much of the global AI industry is converging on increasingly large foundation models, Sakana AI is exploring whether smaller, more adaptive, more context-sensitive systems might ultimately be more powerful in real-world environments.

There is a subtle but important alignment with Japan’s broader innovation culture here. The emphasis is not on domination through scale, but on adaptation within complexity. Systems do not need to be the largest to be effective; they need to be the most responsive to their environment.

Sakana AI is, in many ways, a challenge to the dominant global narrative of artificial intelligence. It suggests that intelligence might be less about brute force computation and more about structured evolution.

Preferred Networks

Origin

Founded in 2014 in Tokyo
Originated from deep learning research communities
Built to bridge academia and industrial application

Activity

AI systems for manufacturing, mobility, healthcare, and science
Works closely with Toyota and industrial partners
Develops full-stack AI from research to deployment

Innovation

Integrates AI directly into industrial production systems
Fuses research and real-world deployment in one loop
Treats AI as industrial infrastructure, not standalone software

If Sakana AI explores alternative definitions of intelligence, Preferred Networks operates at the point where intelligence becomes industrial reality.

Preferred Networks occupies a distinctive position in Japan’s innovation ecosystem. It is neither a pure research lab nor a conventional product company. Instead, it exists in the space between—where advanced machine learning, robotics, and scientific computation are translated directly into industrial systems.

One of the defining characteristics of Preferred Networks is its refusal to separate research from deployment. In many parts of the world, AI research happens in one organisational layer and application in another. In Japan, and particularly within PFN, the two are tightly interwoven.

This creates a different kind of innovation rhythm. Progress is not measured solely in algorithmic breakthroughs, but in whether those breakthroughs can survive contact with the physical world: factories, vehicles, healthcare systems, materials science laboratories.

Its collaboration with major industrial players such as Toyota reflects this philosophy. Rather than positioning AI as a disruptive force external to industry, PFN embeds it inside existing industrial ecosystems, enhancing rather than replacing them.

This is a crucial theme in Japan’s next wave: innovation that is absorbed into industrial structure rather than imposed upon it.

LayerX

Origin

Founded in 2018 in Tokyo
Built by digital-native entrepreneurs
Focused on enterprise workflow transformation

Activity

AI systems for finance, procurement, and corporate operations
Digitisation of legacy Japanese enterprise processes
Integration of AI into existing organisational systems

Innovation

Embeds intelligence into workflows rather than replacing systems
Enables gradual transformation of corporate Japan
Redefines enterprise software as adaptive infrastructure

Not all innovation is visible at the level of physical machines or frontier algorithms. Some of the most important transformations are occurring inside the administrative and organisational fabric of corporations.

LayerX is a case in point.

Its focus is enterprise software, but not in the conventional sense of standalone tools or productivity applications. Instead, LayerX is building systems that embed intelligence into the everyday workflows of large organisations—finance, procurement, compliance, and internal operations.

In Japan, these systems have historically been characterised by complexity, manual processes, and a high degree of paper-based governance. They are deeply embedded in organisational culture, which makes them resistant to abrupt change.

LayerX’s approach reflects a different strategy: rather than attempting to replace these systems, it introduces AI-driven layers that gradually reshape how decisions are made and how work flows through the organisation.

It is not disruption. It is continuous architectural evolution.

This matters because Japan’s corporate sector is vast, structurally important, and deeply interconnected with global supply chains. Even incremental improvements in its operational efficiency can have outsized systemic impact.

LayerX is, in effect, helping to reprogram the internal logic of corporate Japan.

Carbon X

Origin

Founded in Japan as part of climate-tech wave
Built in response to industrial decarbonisation pressure
Emerged from enterprise sustainability needs

Activity

Carbon measurement across supply chains
Industrial emissions tracking systems
Enterprise climate optimisation platforms

Innovation

Treats carbon as a measurable system variable
Embeds climate intelligence into industrial operations
Turns sustainability into a data infrastructure problem

One of the most distinctive aspects of Japan’s innovation culture is the way it reframes global challenges.

Where others often frame climate change in moral or political terms, Japan tends to translate it into a systems engineering problem.

Carbon X exemplifies this approach.

Rather than focusing on consumer-facing sustainability narratives, Carbon X builds infrastructure for measuring and managing carbon emissions across complex industrial supply chains. Its core proposition is that meaningful decarbonisation requires visibility, quantification, and continuous optimisation at the system level.

In other words, carbon is treated not as an abstract goal, but as a measurable variable embedded within production systems, logistics networks, and procurement decisions.

This reflects a deeper Japanese instinct: that complexity is not solved by simplification, but by better measurement and tighter system control.

Carbon X is therefore not just a climate tech company. It is building the informational infrastructure through which industrial decarbonisation becomes operationally executable.

Rapidus

Origin

Established as a public-private semiconductor initiative
Backed by Japanese government and industry leaders
Created to restore advanced chip manufacturing capability

Activity

Development of cutting-edge semiconductor fabrication
Focus on 2nm node manufacturing technology
Collaboration between state, industry, and research institutions

Innovation

Rebuilds national semiconductor sovereignty
Integrates industrial policy with advanced engineering
Attempts frontier manufacturing at global scale

Few companies illustrate Japan’s strategic seriousness more clearly than Rapidus.

Rapidus is attempting something that goes far beyond corporate innovation. It is a coordinated national effort to re-establish Japan’s presence at the frontier of semiconductor manufacturing.

The ambition is to develop advanced 2nm chip production capability—one of the most complex industrial processes in existence today.

What makes Rapidus significant is not only its technical challenge, but its organisational structure. It represents a rare alignment between government, industry, and research institutions, unified around a single industrial objective.

In many ways, Rapidus reflects a different model of innovation governance: not fragmented entrepreneurial experimentation, but coordinated industrial reconstruction.

It is innovation as national infrastructure strategy.

ExaWizards

Origin

Founded in Tokyo in 2016
Built to address societal and demographic challenges
Strong focus on social impact through AI

Activity

AI for healthcare, ageing, energy, and public services
Digital systems for societal infrastructure
Applied AI for complex social systems

Innovation

Uses AI to extend societal capacity
Focuses on demographic and structural challenges
Treats social systems as optimisation problems

ExaWizards operates in a different but equally important space: the application of artificial intelligence to societal systems.

Its work spans healthcare, ageing populations, energy efficiency, and public service optimisation. Japan’s demographic structure makes this particularly significant. With one of the oldest populations in the world, the country faces structural pressures that cannot be solved through traditional labour or productivity models alone.

ExaWizards positions AI not as a replacement for human systems, but as a way of extending societal capacity—supporting healthcare systems, augmenting decision-making, and improving service delivery in contexts where human resources are increasingly constrained.

This is innovation directed not at markets, but at societal continuity.

Abeja

Origin

Founded in Tokyo in 2012
Emerged from applied AI and data science backgrounds
Focused on industrial use cases from inception

Activity

Computer vision and edge AI systems
Industrial monitoring and optimisation
Retail, logistics, and manufacturing intelligence

Innovation

Embeds AI into physical environments in real time
Turns operations into continuously optimised systems
Bridges digital intelligence and physical execution

Abeja represents another key strand of Japan’s innovation trajectory: the embedding of artificial intelligence into physical environments.

Its systems use computer vision and edge AI to monitor and optimise industrial processes in real time. Factories, retail environments, and logistics systems become continuously observable and adjustable.

What is important here is not the AI itself, but its placement—inside the physical flow of production and consumption.

Abeja reflects a broader Japanese pattern: intelligence is not a separate digital layer. It is something that must be embedded directly into operational reality.

Fast Retailing

Origin

Founded in Hiroshima in 1949
Transformed under Tadashi Yanai into global retailer UNIQLO
Evolved from retail store to global system company

Activity

Global apparel design, manufacturing, and retail
Data-driven supply chain and inventory systems
Large-scale retail operations across continents

Innovation

Turns retail into a real-time data system
Integrates design, production, and logistics into one loop
Uses simplicity as a system optimisation strategy

Fast Retailing is often perceived as a retail company. In reality, it is one of the most sophisticated supply chain and data-driven manufacturing systems in the global consumer economy.

Its UNIQLO brand is built on radical simplicity at the product level. But beneath that simplicity lies a highly complex system of global demand sensing, inventory optimisation, and production coordination.

Design decisions are informed by data. Manufacturing is tightly controlled. Distribution is dynamically adjusted across global markets.

Fast Retailing demonstrates a distinctly Japanese form of innovation: reducing visible complexity while increasing systemic sophistication underneath.

Ajinomoto

Origin

Founded in 1909 in Tokyo
Originated in food and amino acid research
Evolved from food manufacturer to biotech company

Activity

Amino acid science and fermentation technologies
Health, nutrition, and biotech applications
Global food and life sciences systems

Innovation

Reframes food as biological system design
Expands into precision nutrition and health science
Converts legacy food expertise into biotech platforms

Ajinomoto illustrates another dimension of Japan’s innovation evolution: the transformation of traditional industries into science-led platforms.

Once known primarily for food products, Ajinomoto is increasingly positioned as a biotechnology company rooted in amino acid science and fermentation systems.

Its expansion into health, nutrition, and life sciences reflects a broader shift: food is no longer seen simply as consumption, but as biological optimisation infrastructure for human health.

This is innovation through scientific reinterpretation of legacy industries.

Characteristics of Japan’s next wave innovators

Across these companies a coherent innovation logic emerges. It is not a style or aesthetic—it is a systemic philosophy of how value is created, scaled, and sustained in complex economies.

1. Systems over Products

In most dominant innovation ecosystems, success is defined by products: a platform, an app, a device, a model, a service. In Japan’s emerging wave, the unit of innovation is fundamentally different.

What these companies build are not products in isolation, but systems that coordinate many moving parts over time.

A robot is not the innovation at Mujin—it is the orchestration layer that allows thousands of robots across different environments to behave intelligently without bespoke programming. An AI model is not the innovation at Sakana AI—it is the underlying architecture for how intelligence itself can be composed, adapted, and evolved. At LayerX, the innovation is not workflow software—it is the gradual reconfiguration of how information, decisions, and approvals flow through entire organisations.

This systems-first mindset reflects a deep engineering heritage in Japan, where complexity is not avoided but embraced and structured. Instead of optimising a single interface or feature, these companies optimise entire value chains, decision loops, and operational ecosystems.

The result is that innovation becomes less visible but more durable. It is not something users interact with directly—it is something that shapes the conditions under which everything else operates.

In this sense, Japan is not building a collection of companies. It is building an interconnected architecture of industrial intelligence.

2. Physical World Intelligence

A striking feature of Japan’s innovation trajectory is its anchoring in the physical world. Unlike digital-first ecosystems where value is often created through software abstraction alone, Japan’s most important innovations remain tightly coupled to physical systems: factories, logistics networks, hospitals, infrastructure, materials, and energy systems.

Mujin’s robots operate in warehouses filled with unpredictable objects and shifting conditions. Abeja’s systems interpret real-time visual data from retail stores and manufacturing lines. Fast Retailing coordinates global supply chains that span manufacturing plants, shipping routes, and retail environments. Even AI companies like Preferred Networks are deeply embedded in industrial and scientific contexts.

This grounding in physical reality matters because it introduces constraints that fundamentally shape innovation. Physical systems are not infinitely scalable or easily replicated. They require reliability, safety, coordination, and resilience.

As a result, Japanese innovation tends to prioritise robustness over speed, adaptability over scale, and operational continuity over rapid iteration.

This creates a different kind of technological output. Instead of fragile systems that perform well in ideal conditions but break under complexity, Japan’s innovation ecosystem produces systems designed to operate under uncertainty, variation, and long time horizons.

In a world increasingly defined by the interaction between digital intelligence and physical infrastructure—autonomous logistics, smart manufacturing, climate systems, energy transition—this grounding becomes a strategic advantage.

Japan is not just building software. It is building intelligence that survives contact with reality.

3. Evolution over Disruption

Perhaps the most culturally distinctive feature of Japan’s innovation model is its approach to change itself.

In many Western narratives of innovation, progress is framed as disruption: new systems replace old ones, incumbents are displaced, and value shifts rapidly from one architecture to another. In Japan, however, innovation is more often framed as accumulation and refinement rather than rupture.

LayerX does not replace enterprise systems—it adds intelligence layers on top of them. ExaWizards does not rebuild healthcare systems—it extends their capacity through AI augmentation. Even Fast Retailing does not reinvent retail each season—it continuously refines a tightly controlled system of design, production, and distribution.

This approach is deeply pragmatic. Japan operates with a large base of mature, highly optimised industrial systems that are too complex and too critical to be rebuilt from scratch. Instead of disruption, innovation must therefore work through integration, compatibility, and gradual transformation.

This creates a different temporal rhythm of innovation. Change is slower in appearance, but often deeper in structural effect. Instead of visible shocks, there is continuous reconfiguration beneath the surface.

Over time, this produces systems that are remarkably stable yet constantly evolving—what might be described as quietly adaptive infrastructures.

The strategic implication is important: Japan’s innovation model is not designed to maximise short-term transformation, but to ensure long-term systemic continuity while still increasing intelligence and capability.

4. Industrial Embeddedness

Another defining feature of Japan’s innovation ecosystem is the degree to which startups and new technologies are embedded within existing industrial structures rather than operating independently of them.

Preferred Networks works closely with Toyota. Rapidus is structurally intertwined with government and major corporate actors. Mujin deploys into global manufacturing and logistics systems that already exist at enormous scale. Ajinomoto evolves from within a century-old industrial base in food science. Even AI companies frequently operate in close partnership with established corporations.

This is not incidental. It reflects a fundamentally different model of innovation diffusion.

In Japan, large corporations are not obstacles to innovation—they are co-architects of it. They provide scale, distribution, operational environments, and long-term investment horizons that startups alone cannot replicate.

This creates a hybrid innovation structure in which startups and incumbents are not in opposition, but in collaboration. Startups bring new technological paradigms; incumbents provide system integration, market access, and industrial depth.

The result is a form of innovation that is less about rapid independence and more about structured interdependence.

This embedded model may appear slower than more fragmented ecosystems, but it allows technologies to scale directly into real-world systems with fewer discontinuities. Innovation does not need to find its way into the economy—it is born inside it.

5. Reliability as Competitive Advantage

In most innovation narratives, success is measured by speed, novelty, or scale. In Japan’s next wave, another dimension is equally important: reliability as a form of value creation.

This is particularly evident in companies like Mujin, Fast Retailing, and Preferred Networks, where systems are designed not just to perform well, but to perform consistently over long periods of time in complex, high-stakes environments.

Reliability in this context is not a passive quality. It is an active engineering objective. Systems are built to minimise failure, anticipate variation, and maintain continuity under stress.

This reflects a deeper cultural and industrial logic. In sectors such as manufacturing, healthcare, infrastructure, and logistics, failure is not merely inconvenient—it is costly, sometimes catastrophic. As a result, trust becomes a core design constraint.

The innovation implication is profound: rather than optimising only for peak performance, Japanese systems often optimise for predictable long-term performance under real-world conditions.

This creates a different kind of competitive advantage. While other systems may scale faster or experiment more aggressively, Japanese systems often win on endurance, integration quality, and operational resilience.

In a global economy increasingly dependent on complex, interconnected systems—autonomous logistics, AI-driven infrastructure, climate systems, healthcare networks—reliability becomes not a conservative constraint, but a strategic asset.

知能化されたシステム経済

Japan’s next wave of innovators is not attempting to win the global technology race on the same terms as others. It is redefining the terms.

Instead of chasing speed, it builds stability. Instead of platforms, it builds systems. Instead of disruption, it builds continuity with intelligence layered into every component of the economy.

Taken together, companies like Mujin, Sakana AI, Preferred Networks, LayerX, Carbon X, Rapidus, ExaWizards, Abeja, Fast Retailing, and Ajinomoto suggest something larger than a startup ecosystem. They suggest the emergence of a new economic model: 知能化されたシステム経済 … meaning, a systems intelligence economy, where the boundaries between software, hardware, biology, and industry dissolve into integrated, adaptive infrastructures.

In a world increasingly defined by volatility, fragmentation, and acceleration, Japan’s quiet approach may turn out to be unexpectedly powerful. Because the most important innovations are not always the ones that move fastest.

They are the ones that become so embedded in the world that the world cannot function without them.

More from the blog