o3 proves that AI has not slowed down, Taiwan Semiconductor becomes the "best choice," and stock prices head towards a historical high
Analysis suggests that the O3 model proves that AI has not slowed down, and without the need to pick winners or technologies, Taiwan Semiconductor is the best way to participate in the AI theme, with reasonable valuation
With the enthusiasm for AI investment unabated, TSMC's stock price is approaching historical highs.
On Tuesday, TSMC's stock price on the Taiwan Stock Exchange rose by 1.4% at one point during the trading session, surpassing the closing record high set in November. Overnight, U.S. chip stocks like NVIDIA surged. Due to investors' continued enthusiasm for AI trading, TSMC's stock price has accumulated an 84% increase this year, likely achieving its best annual performance in 25 years.
It is worth mentioning that TSMC's U.S. stock price premium relative to its Taiwan stock price has risen to over 24%, reaching the highest level since early December, due to discrepancies in performance disclosures between its listings in Taipei and New York.
Analysts indicate that this shift is due to the increased demand for exposure to TSMC's growth in the global market, as this chip manufacturer continues to play a key role in semiconductor and AI supply. Even amid regional volatility, investors have shown strong interest in TSMC's role in advanced chip production for high-growth areas such as AI-driven technologies.
This upward trend highlights the level of attention global investors are paying to companies expected to benefit from the AI boom. TSMC's clients include Apple and AMD, major beneficiaries of the surge in AI spending. The company expects a 36% increase in sales for the December quarter, with a gross margin reaching 58.3%, the highest level since 2022.
TSMC's rise is closely tied to the market's optimistic outlook on the accelerated development of AI.
Previous analysis from the Information Equality WeChat account stated that the narrative around AI has strengthened rather than weakened.
First, it must be acknowledged that o3 has at least proven that AI is not slowing down. The o1 preview to the o1 pro release was only three months apart, and shortly after, we moved directly to o3. Just looking at the "testing time calculation" aspect alone has enormous potential. The issues are clear: high costs and "targeted training mathematicians," but cost and delay are the least important factors. With NV's super-large nodes iterating once a year, costs and delays will certainly rapidly deflate.
Secondly, for investors concerned about catalysts and future expectations, there is an invisible narrative: starting from o3, OpenAI in 2025 will use every method to prove it is infinitely close to "AGI." This will free it from Microsoft's control and renegotiate investment agreements. The previous investment agreement stated that once AGI is achieved, Microsoft would lose its rights to use OpenAI's technology.
Kevin Net, head of Asian equities at Financiere de L Echiquier, stated:
For us, TSMC is the best way to participate in the AI theme without having to pick winners or technologies, and the valuation is reasonable. The next catalysts to watch are NVIDIA's presentation at CES 2025 next month and TSMC's upcoming performance and guidance announcements. **
An earlier article from Wall Street Journal stated that Jensen Huang, the founder of NVIDIA, will deliver a speech at CES 2025 on January 6 at 18:30 Pacific Time, which is 10:30 on January 7 Beijing Time. The RTX 5000 series GPU will be the highlight of the keynote speech