Google and KDDI Want to Back Japan’s AI Startups — But Will This Build Independence or Deepen Gemini Dependence?
Google is expanding its AI Futures Fund strategy into Japan. In an official announcement published on July 28, 2026, the company said it is partnering with Japanese telecom group KDDI to launch the AI Startup Support Program, a new initiative designed to fund and accelerate Japanese startups building AI-native products.
On paper, the program looks like a strong opportunity for Japanese founders: equity investment, early access to Gemini, Nano Banana, Lyria, Google Cloud credits, technical support, business guidance and access to compute resources hosted in Japan through KDDI’s Osaka-Sakai Data Center.
But the strategic question is bigger than startup financing. With this program, Google is not only helping Japanese startups. It is also encouraging the next generation of AI companies in Japan to build around Google AI, Gemini, Google Cloud and Google’s broader model ecosystem.
What is the Google KDDI AI Startup Support Program?
The Google KDDI AI Startup Support Program is a joint initiative from the Google AI Futures Fund and the KDDI Open Innovation Fund. Its goal is to identify, fund and accelerate promising AI-native startups in Japan. According to the official Google AI Futures Fund page, the program combines equity investment, compute credits, technical and business support, and early access to advanced Google DeepMind models.
The program focuses on five areas:
- the future of work,
- knowledge,
- software development,
- creativity and entertainment.
This gives the initiative a broad scope. It is not only targeting infrastructure startups or enterprise software. It also wants to reach founders working on creative AI tools, AI coding products, knowledge platforms and entertainment applications.
Selected startups may receive joint funding from both the Google AI Futures Fund and the KDDI Open Innovation Fund. KDDI’s English program page states that selected startups can receive up to $2 million in co-investment per startup, along with up to $350,000 in credits for Google AI tools and Google Cloud.
This is an important clarification. Google’s blog announcement mentions equity investments but does not specify the amount. The more detailed KDDI program page adds the figure of up to $2 million per startup, which gives founders a clearer idea of the financial scale of the initiative.
What do selected Japanese AI startups receive?
Selected startups can receive four main categories of support.
First, they can receive equity-based funding from both the Google AI Futures Fund and KDDI Open Innovation Fund. This means the program is not a grant or a simple accelerator. Participants are expected to provide an equity share in exchange for capital investment and program resources.
Second, startups get early access to advanced AI models and tools from Google DeepMind, including Gemini, Gemma, Nano Banana and Lyria, according to the official program page. Google’s blog specifically highlights Gemini, Nano Banana and Lyria, while the AI Futures Fund page also mentions Gemma.
Third, the program offers Google Cloud credits and credits for sovereign Gemini and GPU resources provisioned in Japan through KDDI’s Osaka-Sakai Data Center. KDDI says the support includes advanced GPUs available at that data center, while the Japanese page refers to on-premises sovereign Gemini through Gemini on GDC and KDDI GPU Cloud.
Fourth, startups receive hands-on support from Google and KDDI experts, including researchers, engineers, product managers, designers and go-to-market specialists. Google also says selected startups will have opportunities to give feedback directly to the teams building its AI products.
For founders, this combination can be valuable. Access to money is useful. Access to models, infrastructure, technical teams and distribution channels can be even more important.
Why this announcement matters for Japan’s AI ecosystem
Japan has strong industrial, robotics, gaming, telecom and hardware sectors, but it has not yet produced a global consumer AI platform at the scale of ChatGPT, Claude, Gemini or Perplexity. Google’s move into the Japanese startup ecosystem should therefore be read as a strategic attempt to connect local innovation with its global AI stack.
The partnership with KDDI is also significant. KDDI brings local market knowledge, telecom infrastructure, corporate relationships and a venture investment history. Its English page says the KDDI Open Innovation Fund has been active for over 15 years and has invested in more than 175 companies globally.
For Google, this is not just about financial participation. It is about ecosystem formation. If a young Japanese startup builds its first product with Gemini, scales on Google Cloud, receives investment from Google’s AI fund and gets technical feedback from Google teams, it becomes much more likely to remain close to Google’s infrastructure over time.
That is the core strategic logic. Google wants more AI-native companies to build with its models before they become large enough to choose competing infrastructure from OpenAI, Anthropic, Microsoft Azure, AWS, Mistral AI, DeepSeek, Qwen or local Japanese providers.
What Google does not say clearly
The announcement is positive for founders, but it leaves several important questions open.
Google does not disclose the total investment envelope for the program in its blog announcement. KDDI mentions up to $2 million in co-investment per startup, but the number of startups that may be selected is not specified.
The program also does not publish a detailed investment framework. We do not know the average equity share that Google and KDDI may request, how valuations will be negotiated, how much influence investors may have over strategic direction, or whether startups will be encouraged to prioritize Google infrastructure in future product decisions.
The program says participants are required to use Google AI in a material way. It also says this is not an exclusivity requirement and that startups may use other models and tools alongside Google’s AI technologies.
That distinction matters. The program does not formally lock startups into Google only. But in practice, a startup receiving capital, model access, cloud credits, technical mentoring and business support from Google may still become strongly dependent on the Google ecosystem.
The “sovereign Gemini” question
One of the most interesting parts of the announcement is the reference to sovereign Gemini and GPU credits provisioned in Japan through KDDI’s Osaka-Sakai Data Center.
For Japanese companies, this can be important. Local compute infrastructure may help startups serve customers with data residency, latency, compliance or operational requirements. In sectors such as enterprise software, public services, finance, healthcare, manufacturing and regulated industries, where data is processed can matter almost as much as what model is used.
But the word sovereign needs careful interpretation. Local hosting does not automatically mean technological independence. Gemini remains a Google model. Its roadmap, licensing, safety policies, API evolution and commercial terms remain controlled by Google. The sovereignty here appears to be more about infrastructure, deployment and data locality than full independence from foreign AI technology.
For us, this is the central nuance: the program may strengthen Japan’s AI startup ecosystem, but it may also reinforce the role of Google as the infrastructure layer underneath that ecosystem.
Who can benefit from the program?
The most direct beneficiaries are Japanese AI startups working on products in work, knowledge, software development, creativity and entertainment. Startups from pre-seed to Series A appear to be preferred, but the program page says all founders and researchers with a strong idea and clear plan may be considered.
AI coding startups could benefit from early access to Gemini and technical support from Google teams. Creative AI startups may be interested in Nano Banana and Lyria, especially if they are building image, audio, music, media or entertainment products. Enterprise AI startups may benefit from Google Cloud, sovereign compute and KDDI’s local business network.
Developers and technical founders may also gain from direct feedback loops with Google researchers and engineers. That can help startups understand model behavior, optimize costs, test prototypes faster and prepare for future model capabilities before wider market release.
For international investors and AI watchers, the program is worth monitoring because it shows how major AI labs are moving beyond APIs. They are building regional startup pipelines around funding, cloud credits, compute access, local partners and early model access.
The main risk: innovation with a dependency layer
The biggest risk is not that the program is bad for startups. It is that it may create a dependency layer early in the life of those startups.
A founder who receives funding from Google and KDDI, builds around Gemini, uses Google Cloud credits, gets compute from KDDI’s data center and relies on Google product teams for early technical guidance may have fewer incentives to test alternative models or infrastructures later.
This does not mean the program is closed. The official FAQ explicitly says using Google AI is not an exclusivity requirement and that teams can use other models and tools.
But independence is not only a legal question. It is also a technical, financial and operational question. The deeper a startup integrates one AI stack, the more expensive and complex it can become to migrate away from it.
CritiquePlus opinion
CritiquePlus sees the Google KDDI AI Startup Support Program as a smart and strategically important initiative. It gives Japanese AI startups access to capital, models, infrastructure and expertise at a moment when AI-native companies need speed, compute and distribution.
The program is especially interesting because it combines global AI capabilities from Google DeepMind with local infrastructure and market access through KDDI. That makes it more relevant than a generic startup accelerator. It is both a financing program and an ecosystem-building move.
But the critical reading is clear: this is not just support for Japanese innovation. It is also a way for Google to shape the technical foundation of Japan’s next generation of AI startups.
The program should be tested and watched, not blindly celebrated. For founders, the opportunity is real. For Japan’s AI ecosystem, the strategic question remains open: will this help create globally competitive Japanese AI companies, or will it mainly create a new generation of startups built on top of Gemini and Google Cloud?
The answer will depend on the selected startups, the investment terms, the degree of technical freedom, the openness to competing models and whether local infrastructure translates into real strategic autonomy.
What to watch next
The key points to monitor are the number of startups selected, the actual investment amounts, the equity terms, the first use cases, the degree of reliance on Gemini, and whether startups can realistically deploy products using competing models or cloud providers.
Another point to watch is the role of KDDI’s Osaka-Sakai Data Center. If it becomes a serious local AI infrastructure hub, the program could matter beyond startup funding. It could become part of a broader Japanese strategy for AI compute, data residency and enterprise deployment.
For now, the Google KDDI AI Startup Support Program is best understood as a strategic accelerator: useful for founders, valuable for Google, important for Japan, but not a guarantee of technological independence.
Official sources used
Google Blog — official announcement: “Google and KDDI are ready to back Japanese startups”, published on July 28, 2026.
Google AI Futures Fund — official AI Startup Support Program page.
KDDI — official English program page for the AI Startup Support Program by KDDI & Google AI Futures Fund.
KDDI — official Japanese program page, including details on Gemini on GDC and KDDI GPU Cloud.
