Wu Zhiyi, CEO of the Taiwan Semiconductor Industry Association, recently expressed his views on how Taiwan can use its strengths in the server and chip field to overcome obstacles to the development of artificial intelligence (AI) technology. He emphasized that the advancement of semiconductor technology, especially the continuous innovation of chip technology, is a key factor driving the development of generative AI. The following is Wu Zhiyi's insightful analysis of the development of artificial intelligence and the future of Taiwan's technology industry.
The rapid development of artificial intelligence and the breakthrough of semiconductor technology
Wu Zhiyi pointed out that in the past decade, AI technology has undergone a leap from discriminative to generative, and the speed of technological progress is astonishing. The widespread use of generative AI relies on the computing power behind it, which is based on the continuous breakthrough of semiconductor technology.
He further elaborated that with the advent of 2nm chips, technological advances in the semiconductor industry are providing more powerful support for the development of artificial intelligence. Compared to the 10-nanometer chips used a decade ago, the emergence of 2-nanometer chips will greatly improve computing efficiency and provide more support for the complex computing needs of artificial intelligence.
Nine of the world's top 10 companies have joined the R&D and application of AI technology, and AI is expected to contribute more than $15 trillion to the global economy by 2030, a projection that highlights the importance of AI as a core technology for the future.
The rapid rise of artificial intelligence and capital markets
Wu also mentioned that the growth rate of AI start-ups is astonishing. According to the data, the average time it takes for AI start-ups to develop from initial capital to a company with a market value of $1 billion is 3.4 years, which is much lower than the average time for "unicorn" companies in other fields, showing the high return potential of investment in the field of artificial intelligence.
This rapid development trend is closely related to the continuous advancement of semiconductor technology. With the continuous innovation of chip technology, the application scenarios of artificial intelligence have expanded from quality control in semiconductor manufacturing, circuit board printing and packaging to medicine, inventory management, advertising and other fields, showing the penetration and wide application of AI technology in multiple industries.
Figure: Strategic thinking on promoting the development of artificial intelligence in Taiwan's semiconductor industry (Source: Taipei Times).
Energy consumption issues and challenges of energy-saving technologies
Despite the promise of AI technology, Wu also pointed out that the high energy consumption of AI is still one of the main obstacles to the development of this technology. It is estimated that by 2030, the energy demand of AI data centers may account for about 10% of all electricity generation in the United States, which puts huge pressure on the global energy structure and environmental protection.
Therefore, how to develop more efficient data center cooling technology and energy-saving technology will become the key to the future development of the AI industry. Wu Zhiyi stressed that the industry must take into account energy conservation and environmental sustainability while pursuing technological progress, so as to ensure that the long-term development of technology does not cause excessive burden on global resources and the environment.
Challenges and opportunities in Taiwan: Focusing on specialized small language models
As an important base for the global semiconductor industry, Taiwan has encountered challenges in the application and development of AI technology in recent years. Wu Zhiyi pointed out that although Taiwan has significant advantages in semiconductor manufacturing, server and chip design, Taiwan still faces certain shortcomings in the amount of data, computing power and the cultivation of software talents. In particular, Taiwan currently does not have the basic conditions to develop general-purpose large language model (LLM) artificial intelligence. Therefore, Wu Zhiyi suggested that Taiwan should focus on the development of professional small language model algorithms, which will be a key direction for the development of Taiwan's AI industry in the future.
In addition, Wu Zhiyi also mentioned that Chinese mainland accounts for 90% of the production in the global server market, which makes Chinese mainland have a certain leading edge in the field of deep learning integrated circuit technology, unlike companies such as Nvidia and AMD, which mainly focus on the research and development of large graphics processing units (GPUs). Although the theoretical limits of global chip design have not yet been reached, the semiconductor industry will continue to innovate and launch more efficient and powerful computing chips for at least the next 20 years as materials technology continues to advance.
Summary: Taiwan's semiconductors and artificial intelligence are developing synergistically
To sum up, Wu's speech at the conference revealed the key role of Taiwan's semiconductor industry in the development of artificial intelligence technology. He clearly pointed out that breakthroughs in semiconductor technology, especially chip technology, will provide basic support for the development of artificial intelligence technology and help Taiwan occupy a favorable position in the global scientific and technological competition. At the same time, Wu Zhiyi also reminded that although the artificial intelligence industry has broad prospects, the research and development of high energy consumption and data center cooling technologies still need to be paid attention to.
