Why AI Edge will come from Japan’s Startup-Corporate Ties

The term “startup” began gaining widespread use in the 2000s. Before then, companies that created innovative products and services and distinguished themselves in society were called “ventures.” While there’s no unified definition distinguishing ventures from startups, the latter has become frequently used to describe emerging companies that recruit investors based on ideas and leverage the power of internet information dissemination to achieve explosive growth.

A quarter-century has passed since the dawn of the startup era, and significant changes are occurring in the startup ecosystem. On July 15, 2025, during the second day of KOZAROCKS held in Okinawa, Kazuhiko Chuman, Executive Officer and Chief Business Development Officer (CBDO) of Mizuho Financial Group—one of Japan’s three major financial conglomerates—participated in a fireside chat during the opening session. The session was moderated by Masashi Ikeda, CCO of Cohh and Editor-in-Chief of Growthstock Pulse.

Until March of this year, Chuman spent a decade at telecommunications giant KDDI, where he led the KDDI ∞ LABO (KDDI Mugen LABO) program that promotes collaboration between major corporations and startups, and served as head of Corporate Venture Capital (CVC), handling investments and business creation across diverse fields. What drove this figure—often called “the father of Japanese open innovation”—to transition from telecommunications to finance?

The Vision Behind the Career Change

Chuman’s transition to the financial industry represents more than a simple career change. It stems from deep concern about structural challenges facing Japan’s economy and strategic judgment about the historic inflection point of the AI era. His vision is to expand the open innovation framework he built during his KDDI years to broader industries and transform Japan’s industrial structure itself.

The Partner Alliance Experiment

The “Partner Alliance” initiative that Chuman spearheaded during his KDDI tenure is essential to understanding Japan’s open innovation landscape. Typically, when companies engage in business collaboration with startups, they limit their focus to areas that might generate synergies with their own business. As a telecommunications company, KDDI would naturally be expected to attract startups related to telecom business. However, in reality, startups from diverse fields gathered at KDDI ∞ LABO.

The program started in 2011, a few years after smartphones began emerging. It was an era when everyone was looking for ‘trendy apps.’ As smartphones proliferated, people were searching for interesting games and such, so we began recruiting entrepreneurs in those areas. Originally, KDDI was doing this to support startups.

But after about three years, as our business scope expanded, we saw growth in services like real estate matching, ride-sharing matching, and hotel matching. These services were somewhat distant from a telecommunications company.

Meanwhile, we often heard from major corporations that they wanted to co-create businesses with startups but didn’t know how and asked us to teach them. If companies beyond KDDI—real estate companies, railway companies, TV stations, and other corporations—could participate, it would broaden options for startups. So we invited corporations that wanted to support startups to join us, creating opportunities to gather about once a month. (Chuman)

The domestic corporate partner alliance that began in 2014 reached 100 participating companies by July 2024, a decade later. The Partner Alliance’s startup business support in fiscal 2023 totaled approximately 1,000 cases annually, making it a major hub for open innovation in Japan.

The Challenge of Industrial Structural Reform

But why did Chuman decide to transition to the banking industry at this particular moment? The background reveals strong awareness of structural challenges facing Japan’s entire economy.

Since the early 1990s bubble collapse, Japan has experienced prolonged economic stagnation for about 30 years, a period known as the Lost Decades. The failure to establish an entrepreneurial culture and the formation of an environment where startups struggle to grow has put Japan significantly behind Western nations and other Asian countries.

As this era of stagnation finally comes to an end and the world searches for new growth engines, generative AI startups led by ChatGPT emerged from 2022 onward.

Japan has been continuously called the Lost Decades, and we’re actually facing the problem of how to break through this. In a situation where investment in completely new business models and technologies has disappeared, I wonder if we can really create a bright future for Japan by continuing to focus only on AI efficiency improvements of existing services.

In this regard, banks are involved in all kinds of businesses and, while not operating specific businesses themselves, can use finance to change industry structures, create new models, and support them. In a telecom company, investments tend to center on areas related to our own business domain, but in the banking sector, we can engage with much broader industries.

We’re leveraging this position to set an extremely ambitious goal of transforming the industrial structure itself. Over the next decade, we want to commit to creating new mechanisms for Japan. Rather than mere efficiency improvements or existing system enhancements, we want to open new possibilities for Japan through fundamental transformation of industrial structures. (Chuman)

How AI is Changing Startup Conventions

The first half of 2025 saw dramatic changes in the AI field. NVIDIA’s market capitalization reached 4 trillion US dollars, demonstrating how high investor expectations have become for the AI sector. The most notable aspect of Chuman’s session was his perspective on the structural characteristics of AI startups. AI startups are changing the fundamental assumptions of the entire ecosystem, going beyond technological advancement.

Computing Resources

Traditional internet companies, regardless of their growth speed, typically started with relatively small initial investments and gradually scaled up. However, the AI era has fundamentally overturned this conventional wisdom.

The first characteristic facing AI startups is enormous demand for computing resources. According to Chuman, AI development requires high-performance GPUs, necessitating investments of hundreds of millions of yen annually in computing resources from the founding stage. This reality is dramatically changing startup fundraising strategies.

Traditional startups could begin operations with minimal investment if they had ideas and programming skills. However, for AI startups, access to high-performance GPU clusters determines business success or failure. Developing large language models or image generation AI often requires computational costs in the millions of dollars.

Data Access Challenges

The second characteristic facing AI startups is the critical importance of data access. AI algorithms demonstrate practical value only by learning from massive amounts of high-quality data, but it’s extremely difficult for newly founded startups to prepare necessary data independently.

Valuable data accumulated across industries is held by existing companies, making access extremely difficult for new startups. In healthcare, medical image data needed for diagnostic support AI; in finance, transaction data; in manufacturing, production data—all are strictly managed from privacy and regulatory perspectives.

This reality has made the traditional “prototype-first” approach obsolete. Instead, a new model is needed where startups form partnerships with large corporations and government entities from the founding stage, building businesses while receiving data access.

Structural Changes in the Investment Environment

The arrival of the AI era has brought fundamental changes to startup fundraising environments. These changes expose the limitations of traditional venture capital (VC)-centered investment ecosystems while highlighting the need for new investment models.

Venture Capital Limitations

Venture capital firms must raise fund capital from institutional investors like pension funds and corporations, then recover investments within a certain period, returning them with added profits. Typically, fund redemption periods are set at 10 years, requiring portfolio company exits (IPOs or M&As) within this timeframe.

Particularly in deep tech and AI fields, where technology development to commercialization often takes over 10 years, compatibility with traditional VC models is limited. Projects requiring extended periods from basic research through prototype development, proof of concept, regulatory compliance, and market entry often don’t suit typical VC fund investment criteria.

The Rise of Corporate Investment

Conversely, corporate investment can maintain more flexible, long-term perspectives. Since corporations invest their own capital, they have no redemption obligations to external investors. This enables long-term investment decisions and support for startups requiring extended technology development periods.

More importantly, corporations serve not merely as capital providers but as strategic partners. They can provide assets critically important to AI startups: data provision, customer base sharing, access to computational resources, and regulatory compliance support.

Many corporations aren’t just pursuing quarterly profit accumulation. They have corporate philosophies and founding spirits like ‘we started this company to bring this to society.’ As a result, they find it easier to support people who align with these values, seeking to support entrepreneurs’ visions rather than short-term profits. (Chuman)

The Swing-by IPO Strategy

The “Swing-by IPO” strategy that Chuman has long advocated becomes even more important in the AI era. Named after the “swing-by” navigation technique used by spacecraft to accelerate using gravitational pull, this strategy involves startups leveraging large corporate assets to accelerate growth before pursuing IPOs.

Actual examples of Chuman’s involvement include Soracom and Elyza. Soracom, founded in 2014, became a KDDI subsidiary in 2017 and achieved IPO on the Tokyo Stock Exchange Growth Market in March 2024, becoming Japan’s first “swing-by IPO” case. ELYZA became a KDDI Group subsidiary in 2024, advancing LLM development and social implementation.

Instead of startups growing alone over five years, they enter large corporate groups, use those assets to dramatically accelerate growth speed and become unicorns. Joining a group isn’t the end—they aim for public listing again and continue growing from there.

I make a distinction between what I call ‘business creators’ and typical ‘entrepreneurs.’ I have no interest in people who just want to start companies, go public, and make money. Business creators have convictions like ‘I want to solve societal problems’ or ‘I want to do this particular thing,’ and to realize these goals, they need greater influence.

True business creators should choose the most efficient means to create the biggest impact on solving social issues, and they should explore the shortest route to scale up, regardless of method. This thinking is the core philosophy behind the swing-by IPO strategy. (Chuman)

Connected Individuals, Divided World

The contemporary world experiences two seemingly contradictory phenomena simultaneously. At the individual level, cultural integration advances through internet and global platform proliferation, while at national and regional levels, political and economic divisions deepen.

The Gap Between Culture and Politics

Internet and global platform proliferation has normalized cross-border distribution of content and products. The phenomenon of Korean drama “Squid Game” becoming a worldwide topic through Netflix, and Japanese anime and VTubers gaining fans across borders, symbolizes this change.

Previously, each country’s content was limited to local markets due to language barriers and distribution constraints. However, with the emergence of global platforms like Netflix and Spotify, excellent content can now be distributed worldwide instantly.

Meanwhile, political and ideological divisions are intensifying. US-China confrontation, the Russia-Ukraine war, Middle East instability—international society is fragmenting into multiple blocs. This division significantly impacts economic activities, with global supply chain reconstruction and “decoupling” in technology fields progressing.

Things that can expand globally through platforms have emerged, alongside things that absolutely cannot compete globally and must operate regionally. Among startups, some companies want global strategies while others emphasize regional strategies. Though seemingly contradictory, both are correct approaches, so each company must understand its business characteristics and choose appropriate strategies.

Entertainment content and fashion brands can use global platforms to target the entire world as their market, while financial services and government-related businesses often require regional strategies due to each country’s regulations and political constraints. Companies need to determine which nature their business possesses and select corresponding strategies. (Chuman)

Japan’s Cultural Adaptability and Glocal Strategy

Throughout history, Japan has demonstrated unique cultural adaptability. The ability to actively adopt foreign cultures while transforming them into distinctly Japanese values, rather than mere imitation, could become Japan’s greatest weapon in the AI era.

Historical Pathways to AI Application

Historically, Japan has created original cultures while actively adopting foreign influences. Sushi and ramen exemplify this—particularly ramen, which despite Chinese origins, developed into a completely different global product. Even today, the fact that Japanese Italian cuisine receives higher evaluation than in Italy itself, and that Japan has the world’s most Michelin stars, demonstrates this ability to improve foreign cultures and create new value.

This historical cultural adaptability has great potential for the AI era. Chuman analyzed that Japan’s strength lies in its ability to interpret global developments, localize them, and create original solutions.

In AI, global foundational technologies have already been established. Western companies like OpenAI, Google, and Anthropic are leading. Since it’s difficult for Japan to enter these areas now, I think we should focus on applications—what kind of localized solutions we can create. (Chuman)

While Japan lags in foundational technologies, this perspective suggests that in application areas adapting these technologies to Japan-specific social challenges, the historically proven cultural adaptability can enable unique value creation. This capability becomes a powerful weapon in developing AI solutions for challenges Japan faces: aging society responses, disaster countermeasures, and technological implementation of omotenashi (hospitality) culture.

Diversity Value Seen in Okinawa

Reflecting on the event’s Okinawa location, Chuman emphasized the importance of regional cultural diversity.

Issues like depopulation, regional problems, and remote island challenges wouldn’t work if taken to Silicon Valley, where typical global technologies emerge. Those needs simply don’t exist there. But regions worldwide face similar challenges to Japan. We should take services born in Japan to such regions and create partnerships. I think this is what ‘glocal’ truly means. (Chuman)

Chuman also noted that not everyone needs to aim for Google or Amazon, and that we must break free from the uniform definition of first succeeding in Japan, then America. His proposal to consider Asian markets from the start and diversify definitions of global success accurately captures problems in traditional startup ecosystems.

Success Beyond Entrepreneurship

The human resources required in the AI era also differ significantly from what traditional startup ecosystems envisioned. Not everyone needs to become entrepreneurs; rather, developing talent with diverse specializations who support the entire ecosystem is crucial.

Chuman, who teaches entrepreneurship courses at the University of Tokyo, spoke candidly about current entrepreneurship education challenges. Interestingly, among students he’s taught over six years, while some have raised funds from investors, very few have achieved significant business growth. Conversely, students who don’t attend classes produce entrepreneurs who succeed in significantly growing their businesses. When asked why they don’t participate in courses, they answer, “We’re too busy with our businesses to attend classes.”

From this experience, Chuman gained important insights about entrepreneurship education goals. Not everyone needs to become entrepreneurs. Since excellent entrepreneurs work in teams, we need team members who support these entrepreneurs with entrepreneurship knowledge, global perspectives, and understanding of regional and Japanese strengths, capable of creating unique value systems.

A Message for Ecosystem Transformation

Toward the session’s end, summarizing the discussion, Chuman delivered a powerful message to participants.

There are moments when the world definitely changes. Now is exactly that time, and since winners haven’t been determined yet, there’s no need to give up thinking ‘it’s hopeless anyway.’ Rather, everyone has opportunities to challenge. Why don’t we all try non-continuous challenges, not extensions of the past? That’s how Japanese society will change.

These words convey strong conviction that at the historic inflection point of the AI era, challenges involving fundamental transformation rather than incremental improvements are necessary. Particularly in the AI era, possibilities exist for challenges in new domains where existing winners haven’t been determined.

Chuman’s proposals provide important guidance for creating Japan-originated global companies in the AI era. For him, having continuously explored ways to support startups, the financial industry represents a new experimental field. While the experimental stage has changed, his exploratory spirit for creating better ecosystems remains unchanged.

Standing at the major inflection point of the AI era, Chuman’s message resonates throughout Japan’s startup ecosystem. Transition from the traditional “start small and gradually expand” model to a “start big with partnerships from the beginning” model. From VC-centered investment environments to the growing importance of strategic corporate investment. And from global concentration to glocal strategies.

This session, held in Koza—a town adjacent to the US Kadena Air Base where Japanese and American cultures blend—truly opened doors to the new possibilities of the Japanese startup ecosystem. The applause concluding the session conveyed participants’ expectations and empathy for Chuman’s proposals. Japan’s reversal strategy for the AI era is beginning to move right now.