The three giants of AI rarely call for slowdown. Legal person: The advantages outweigh the disadvantages, and the mid- to long-term fundamentals remain unchanged.
Anthropic CEO Amodai called for slowing down the development of cutting-edge models, and Musk and Altman successively supported it; legal persons believe that moderately lengthening the iteration cycle will be conducive to enterprise introduction.
Quick Look
- Anthropic CEO Amodai called on the global AI industry to speed up the development of cutting-edge models and strengthen safety assessments, and Musk and Altman successively expressed support.
- The legal person believes that this move has more advantages than disadvantages.
- Moderately lengthening the iteration cycle will help companies accumulate introduction cases, and major CSP manufacturers continue to invest in infrastructure construction.
AI-generated summary
Why It Matters
The heads of AI giants such as Anthropic, xAI and OpenAI have successively called for slowing down the development of cutting-edge models, causing the market to pay attention to the pace of the AI industry.
The AI industry has been making waves again recently. Anthropic CEO Dario Amodei called on the global AI industry to slow down the development of cutting-edge models and strengthen safety assessments. Elon Musk, the head of xAI, and Sam Altman, CEO of OpenAI, also expressed support. The "AI Big Three" rarely called for slowdown, which attracted market attention. However, legal persons believe that the benefits outweigh the disadvantages, and the mid- to long-term fundamentals of AI remain unchanged.
Lin Yijun, fund manager of Nomura Taiwan New Technology 50 ETF (00935), said that AI giants have recently stated that they will slow down the development of cutting-edge models. In addition, the U.S. Congress and local governments have paid more attention to issues such as AI security and data center power and water consumption. Market doubts may be amplified in the short term, but the actual cost of slowing down model iteration may be lower than outsiders imagine. She believes that enterprises have not yet fully digested the application potential of existing models. Properly lengthening the iteration cycle will be conducive to the accumulation of imported cases, and allow enterprise customers to more flexibly plan budgets and AI deployments.
Lin Yijun pointed out that even though major CSP companies face debt costs of about 7% to 8%, they continue to invest in AI infrastructure through bond issuance or their own funds, reflecting the considerable commercial interests behind them. As 2028 approaches and data centers begin to contribute revenue, market doubts about capital expenditures are expected to fade, and the focus will return to profits and cash flow growth.
Du Xinpei, manager of the Yushan Small and Medium Cap Fund, believes that the current Taiwan stock market is in a "weak macroeconomic, strong industrial profit" pattern. The stock market has insufficient trading momentum and will remain range-bound until the fourth quarter. Based on the Taiwan stock price-to-earnings ratio of 17 to 18 times, the market index is estimated to fall at 43,000 points. The range to 49,000 points is more likely, but the good news is that in the second quarter, the profits of Taiwanese companies as a whole continued to deliver outstanding results. In particular, the single-quarter earnings growth rate of listed companies reached approximately 110%, and the after-tax net income reached a record high of 1.8 trillion yuan, reflecting the intensity of demand for AI that exceeded expectations.
What to Watch
AI outlook — possibilities, not facts
Taiwan stocks maintained a range-bound pattern in the fourth quarter
Likely · Within months
Open Questions
- What is the specific timetable for major AI companies to actually slow down model development?
- What is the actual effectiveness of enterprises digesting existing AI models?





