Will the U.S. restrict AI data centers impact Taiwan? Du Yijin: Can block construction, but cannot block demand

After the U.S. midterm elections, multi-state AI data center policies may be adjusted. Wall Street investors sold related stocks. In this regard, Du Yijin, founder of Taiwan Artificial Intelligence Laboratory, said that as long as AI continues to create real value, Taiwan will be the best partner in the world. Demand will not disappear after the US election, it will just exist in different ways.

Du Yijin said in a Facebook post that data centers can be banned, but the demand for AI cannot be stopped. With the recent U.S. midterm elections, Wall Street has begun to question the huge investment in AI data centers. Power shortages, land disputes, environmental impacts, local government opposition, and even calls to restrict or suspend the construction of data centers. The market is beginning to worry that the hundreds of billions of dollars invested by technology giants in AI infrastructure will eventually become a bubble?

Please read on...

Du Yijin pointed out that today's speech coincided with the global stock market crash, and someone asked a related question. Du Yijin asked a question, "If the government prohibits the construction of new data centers, will you stop using AI?" This question is like many people who don't like base stations being built next to their homes, but it is impossible for their mobile phones to have no signal. Data centers provide computing power, and AI is a hard demand. Restricting supply does not mean that demand disappears. This is the most important key to understanding the next wave of AI industry changes.

Du Yijin continued that the ban on building power plants does not mean that humans do not need electricity. More than a hundred years ago, electricity began to change human society. Factories need electricity, households need electricity, and cities need electricity. Even if a city refuses to build a power plant due to pollution, land or safety issues, it does not mean that the city will no longer need electricity. Power plants continue to improve their impact on the environment and surrounding areas, move to other places, transmission lines continue to expand, and power generation technology continues to advance. Electricity has gradually become an indispensable infrastructure for modern civilization.

Du Yijin analyzed that AI is going through a similar history. When AI begins to participate in software development, medical diagnosis, financial analysis, scientific research, industrial manufacturing, national defense security and daily decision-making, computing power is no longer just an investment project of technology companies, but a part of the productivity of the entire society. You can object to the location of the data center for various reasons such as energy and environmental protection, but it is difficult to ask yourself to give up using ChatGPT, hospitals to give up AI diagnostic tools, and scientists to give up AI research capabilities.

Du Yijin added that if data centers are not built, demand will find outlets. If a certain state in the United States restricts the construction of data centers, what will happen? Businesses may move to other states, looking for places with lower electricity prices, sufficient grid capacity, and easy access to land. If the overall construction costs in the United States increase, some investment may be transferred to Canada, Europe, the Middle East or Asia. If large-scale centralized data centers face restrictions, investment in federal distributed computing, edge AI, internal enterprise computing power, and more efficient chip architecture may be accelerated. The supply method is restricted, which is completely different from the AI ​​industry losing demand.

Du Yijin said that Taiwan's key role is increasing. The development of AI in the next ten years will not only involve model parameters, the number of GPUs, or chip manufacturing processes. The deeper competition is the overall capabilities of energy, power grids, advanced semiconductors, memory, cooling, networks, and data management. Countries with advanced manufacturing processes that can provide energy-efficient, trustworthy chip supply chains and secure computing environments will gain an important strategic position in the AI ​​era. This is also an opportunity that Taiwan cannot ignore.

Du Yijin bluntly said that Taiwan not only has the world's key advanced semiconductor manufacturing capabilities, but also plays an important role in AI servers, memory-related supply chains, network equipment, power management and cooling technology. As long as AI continues to create real value, Taiwan will be the best partner in the world. Demand will not disappear after the US election, it will just exist in different ways.