Today, I’d like to explore what 2025 means for the business of LLMs (Large Language Models). A recent topic of discussion is that the models that have succeeded in the existing IT industry do not necessarily apply in the world of LLMs. I see this situation as an opportunity to pave new paths.
News articles have been discussing the concept of “first-mover disadvantage,” where early success can turn into a liability. The reasons are clear: the immense cost of computational resources, labor expenses, and the rapid pace of evolution. This has created an environment where later entrants can leverage open-source technologies to compete at a lower cost. To me, this is evidence that the industry is entering a transitional phase.
Our project prioritizes not just technological advancement but how technology can be seamlessly integrated into society. Given the challenges of the LLM business in 2025, our focus should be on “unique differentiation strategies” and “integration with software.” For example, developing AI models specialized in specific industries or providing customized solutions tailored to each company will be key to building a sustainable business model.
What I particularly want to emphasize is the importance of “human judgment” in utilizing this technology. Clearly distinguishing what AI can handle and where human intervention is necessary will be a decisive factor in business success. Building systems where technology and humans coexist while leveraging regional characteristics and social contexts—this is the direction our project aims to take.
2025 is not just a year of technological competition. It is a year to establish sustainable revenue models and test how technology and human society can coexist harmoniously. In this period that demands a re-evaluation of business models, what steps should we take? How would you navigate this challenge?