The article explains the trend toward larger units used to express figures in data-center investment and semiconductor manufacturing. The IMF estimates total AI investment at $1.4 trillion, while GlobalFoundries' investment in a new US factory amounts to $21 billion.
Reports of large-scale data-center investments have increased this year, highlighting how the units used to express total investment now appear larger than before. According to materials released by the IMF, total investment by AI companies, including internal funds, equity and corporate bonds, is expected to reach $1.4 trillion (approximately ¥220 trillion). Against the backdrop of investment and financing trends in Japan and the United States, the article aims to provide a sense of the scale of data-center and AI investment.
Large projects are also increasing in the semiconductor sector. As part of Japan's investment and financing in the United States, GlobalFoundries is reported to be planning a new factory in New York State with an investment of up to $21 billion (approximately ¥3.3 trillion). The article organizes changes in the units used to express data-center investment and supercomputer performance and considers how enormous figures are represented today.
The article reviews the history of indicators for data-center investment and supercomputer performance while explaining changes in the units used for semiconductor miniaturization. The current leading edge is the 2nm rule, and design rules have shifted from μ (microns) to nm and then to the atomic-scale Å. It also notes that NVIDIA's Vera Rubin integrates approximately 6 trillion transistors. These developments illustrate the expansion of global investment and the rapid evolution of semiconductor technology.