Is AI an Opportunity or a Crisis for Japan?~Part 1 (Recognizing the Crisis)

Jim Zemlin, President of the Linux Foundation (LF), visited Japan on March 16, just before his trip to South Korea, and we had the opportunity to exchange views. Although the schedule was tight, we were also able to arrange meetings with government officials. This fact alone demonstrates that, under the current administration, AI is positioned at the core of national strategy, alongside semiconductors and space.
The discussion covered a wide range of topics, including generative AI, AI for Science, and physical AI, but what I found particularly significant was the progress being made in Agentic AI. While “generative AI” is a technology that unilaterally “provides answers,” “Agentic AI” performs tasks by receiving feedback on results and improving itself. In other words, AI functions not merely as an intelligent output device, but as an “entity that continuously operates” intelligently. Furthermore, AI is evolving from a support tool into an agent that takes action.
A symbolic example of this shift is the rapid expansion of the Agentic AI Foundation (AAIF), led by LF. The launch of the AAIF was announced by Zemlin, the CEO, on December 10 of last year in both Tokyo and San Francisco. On that same day, he made the announcement in front of over 700 attendees in Tokyo, then boarded a flight to San Francisco to announce the launch to an audience of over 2,000 people. And just three months after its launch, approximately 170 major global tech companies had joined. It is reported that discussions on standardization and protocol design are already in full swing within the AAIF.
What is important here is that these discussions are not at the research stage, but are premised on implementation. In other words, Agentic AI is no longer just a concept; it has entered the design phase as an industrial infrastructure. The speed at which this is happening is astonishing. It took over a decade for the internet to become widespread, and cloud computing similarly required a long time to take hold. However, with Agentic AI, standardization and implementation are proceeding simultaneously, and adoption in corporate settings is advancing rapidly. With Agentic AI, we are no longer at the “consideration stage”; the question now is “whether or not to participate in implementation.”
Behind this lies the dramatic reduction in software costs brought about by advances in AI coding. As a result, companies are now able to dive straight into implementation, accelerating the pace of transformation by several to dozens of times compared to the past. Consequently, U.S. companies have already begun restructuring their workforce, laying off thousands to tens of thousands of coding-focused employees. However, this is not merely a wave of layoffs; these companies are also rapidly hiring new architects and orchestrators. A corporate transformation has begun.
What must not be overlooked here is that this change is not merely the adoption of AI, but rather AITX (AI Transformation). This is not a matter of simply enhancing existing operations, but a structural transformation that redesigns operations themselves with AI as a fundamental premise. And the most critical factor is Japan’s position. At present, very few Japanese companies have established a significant presence at the core of standardization. If this continues, there is a risk that, as in the past, Japan will once again find itself in a situation where it possesses the technology but is unable to participate in shaping the rules. Moreover, this time, catching up later will be extremely difficult.
In conclusion, the question is not whether to adopt AI. It is a choice between participating in the design of this new industrial infrastructure or remaining merely a user. Delaying this decision could lead to an irreversible gap.