Federal Reserve Interest Rate Hikes and the Theory of AI Speed Adjustment... Power and Data Center Infrastructure are Key

In conjunction with the interest rate hike stance of the Federal Reserve System, voices are emerging that global big tech companies must adjust the speed of AI development. While some analyze the cause as the rise in capital procurement costs due to interest rate hikes, the limitations of the essential infrastructure required to drive AI are pointed out as a more fundamental variable.

Correlation between Interest Rate Hikes and the Theory of AI Speed Adjustment

According to a video by Economics Master Gwak Su-jong, it is noteworthy that the timing when some CEOs of the five major hyperscaler companies mentioned the need for AI speed adjustment coincides with the timing of the Federal Reserve System's interest rate hikes. As the Federal Reserve System began to make the dollar more expensive by raising interest rates from 3.75% to 4.0%, the burden of capital procurement for companies increased. This is because it is possible to infer that the rising "cost of money" for companies and the difficulty in capital procurement due to interest rate hikes are the causes of the speed adjustment.

The speaker explained that while this analysis is "not wrong, it is also not entirely correct." When a single event occurs, there is not just one single cause that triggers it, as multiple reasons act in combination to satisfy the justification. Interest rate hikes are not limited simply to economic policy. High interest rates mean the dollar becomes stronger, which is a complex factor that also acts on the political and diplomatic influence of the United States by strengthening the status of the dollar as a reserve currency.

Mentioning the change in the United States' economic position, Dr. Gwak explained that unlike in 1945 after the end of World War II, when the nominal GDP of the United States exceeded 60% of the world's GDP, it has now decreased to about the 25% level. Amidst these macroeconomic flows, such as the change in economic power and the hegemony competition between the United States and China, the analysis suggests that the ripple effects of interest rate policy on future industries like the AI industry must be examined comprehensively.

The Pyramid Structure of the AI Industry and the Importance of Infrastructure

To understand the development pattern of the AI industry, it is necessary to look at the 'pyramid structure.' It is the same logic as how the invention of the airplane by the Wright brothers in 1903 required various infrastructure industries, such as materials, components, metals, chemicals, and refining technologies, to develop simultaneously to lead to the aviation industry. AI also sits at the apex of the process where software and hardware combine to evolve into 'smartware.'

The speaker identified power and data centers as the core key industries that must support the bottom to produce the result called AI. This is similar to how the automobile industry must distill crude oil to obtain fuel and be supported by engines, motors, and precise metal performance to move. For AI to operate, the following stepwise infrastructure must precede it:

• Power: The fundamental power source that moves AI (playing a role like fuel for a car)
• Data Centers: Infrastructure that collects, processes, and transmits information through many central computers and sensors
• Platform Companies: The entities, such as Meta, Alphabet, Microsoft, and Tesla, that produce results based on data centers

Ultimately, the technical achievements such as Generative AI or Large Language Models (LLM) produced by platform companies depend on how stably the power supply and data center infrastructure at the bottom are established. The video mentioned the issue of power as a power source required for AI to operate, suggesting that these infrastructural limitations are the key factors determining the speed of the industry.

The Future of the AI Industry and Technological Diversification

Currently, the AI industry is developing around the GPU, but there is a high possibility that it will diversify according to future technical requirements. The video explained that while the GPU is suitable for Generative AI and LLM, the demand for NPU or TPU may be higher in specific industrial fields such as B2B. In other words, there is a prospect that AI technology itself may become multipolar depending on its use.

In conclusion, the 'speed adjustment' mentioned by AI companies is not limited simply to cost issues caused by interest rates. The analysis is that physical limitations, such as the problem of power supply to drive AI and the speed of constructing the data center infrastructure to support it, act as fundamental variables determining the expansion speed of the industry.