Who Will Build the AI Infrastructure?

Hashimoto Gumi site. A dialogue between top leaders from different fields: a construction company and an AI company (From left: Taku Toguchi, CEO of AI inside Inc., and Masanori Hashimoto, CEO of Hashimoto Gumi Co., Ltd.). Hashimoto Gumi Co., Ltd. (Headquarters: Yaizu City, Shizuoka Prefecture; Representative Director/CEO: Masanori Hashimoto) and AI inside Inc. (Headquarters: Minato-ku, Tokyo; Representative Director/CEO: Taku Toguchi) held a dialogue between the heads of both companies under the theme of 'Viewing AI as Social Infrastructure'. Is AI merely an operational efficiency tool? Or will it evolve into an infrastructure that supports society, like electricity, roads, and waterworks? The dialogue featured Mr. Hashimoto, Representative Director and CEO of Hashimoto Gumi Co., Ltd., known as a 'leading DX-utilizing regional general contractor,' and Mr. Toguchi, Representative Director and CEO of AI inside Inc., leading an AI company from Japan. From the seemingly distant fields of construction and AI, candid opinions were exchanged on 'AI as social infrastructure,' 'AI sovereignty in Japan,' and 'Community-based AI utilization.' For dialogue details, click here. Why does a construction company talk about AI? - An infrastructure philosophy that 'building a box is not the end'. 'It might seem surprising for a construction company to talk about AI,' CEO Hashimoto prefaced, before stating: 'What we handle are tangible social capital like roads, bridges, and buildings. However, infrastructure is not established with those alone. It only becomes social infrastructure when it includes the ground, water, electricity, and operational mechanisms.' Although not usually noticed, society stops functioning the moment it disappears. The infrastructure that Hashimoto Gumi has faced is not just about the moment of completion, but rather something premised on being used for decades, sometimes up to 100 years. From this perspective, AI is also not just 'something used because it's convenient,' but needs to be reconsidered as an entity—an infrastructure—where society struggles if it stops. The perspective of social common capital - Visible infrastructure and invisible infrastructure. The dialogue also touched upon the concept of 'social common capital.' Not only roads and buildings, but also fields like finance, education, and institutions are foundations that support society. 'If the financial system stops, daily life cannot be sustained. I believe infrastructure is not about whether it has a physical form, but whether it is essential for society,' (CEO Hashimoto). CEO Toguchi agrees. 'AI is not infrastructure as long as it is used as a convenient tool. I think it only becomes infrastructure when it reaches a stage where society cannot run if it stops.' Can AI become social common capital? This question became the core axis throughout the entire dialogue. 'Can AI become infrastructure?' - Top leaders from construction and AI deepen the discussion. How to connect the decisively different 'time axes' of construction and AI. One major difference between the construction industry and the AI industry is the time axis. 'Buildings are used for decades, sometimes 100 years. On the other hand, AI and GPUs become obsolete in a few years, and models become outdated in half a year. However, when operating large-scale AI, it ultimately comes down to power, semiconductors, and data centers,' (CEO Toguchi). CEO Hashimoto responds while incorporating his field sense related to data center construction. 'Current data centers consume massive amounts of power, but it's hard to imagine no innovation occurring in this field. What is important is whether the logic and philosophy supporting its contents can endure as infrastructure.' Fast-changing AI and infrastructure designed with a long-term perspective. How to connect these two is precisely the question being asked now. AI implementation that started from the field - Practices of Hashimoto Gumi, an advanced DX regional general contractor. AI utilization at Hashimoto Gumi started not as a trendy DX, but from solving issues on the ground. At the construction site of the Osaka-Kansai Expo, a problem arose where drawings written in foreign languages could not be understood. Therefore, they converted the drawings to PDFs, fed them into generative AI, and had it translate and organize the content. As a result, work that traditionally took about a week was shortened to about 15 minutes, significantly reducing the workload on site. 'Leave the work people don't need to do to AI. In turn, people can focus on the work they originally should be doing,' (CEO Hashimoto). At construction sites, tablets and AI are being utilized to streamline drawing confirmation and information sharing. The Yaizu Model - Making AI a 'regional foundation'. In the latter half of the dialogue, the 'Yaizu Model' envisioned by Hashimoto Gumi...