A Quick Note of AI (12-2025)
- Jia Han
- Dec 23, 2025
- 3 min read
Recent development in AI calls for a review. For the start, there is some talk of a stock market crash. The book 1929 and so on. I do not give much credence to that [1].
Another is Oracle and OpenAI circular sales. This is a bit concerning. This could cause problems if their business plan does not pan out. OpenAI perhaps has to do this. OpenAI and Google are in a ferocious battle [2-4]. If OpenAI loses, it could mean its death, or at least a secondary player. The market might allow two or four players (Google, Microsoft, Amazon, Meta). Gemini 3 is reported to be much better than ChatGpt 5.0. But do not rely on a snapshot because both will try to improve and the leader may go back and forth several times before the final winner is determined. Do I worry about OpenAI or Oracle? I have some Oracle but not too much. According to Gene Munster, OpenAI is OK [5]. He is usually quite good, cool and objective.
Wedbush’s Sherlund comments [6] are good. A key is to design some architecture(s) that can reduce computer power required at the same time produces good results. This is uncertain at this moment but could decide the winners and losers. The All In podcast is very good; keep an eye on it. The other channels are CNBC, FOX, Boomooberg. BTW, I would not worry about deepseek. Deepseek uses learned data as input, which is no good for a serious user. Plus, does anyone think that a big company or sovereign will use deepseek? Just in order to save a few thousand dollars? Such a person must be insane.
Palantir is another to pay attention. I heard about it for sometime but the stock is always too expensive (by the ordinary standard). During last April’s market selloff, I bought some but too little. The reason to pay attention is that Palantir is the only company so far that is able to apply AI to general applications and makes money. Hyperscalers, Meta in particular, are able to make internal works more productive. AI can be applied to medical research (I have written one) and banks (Jamie Dimon, Larry Fink), and legal firms. Most of these applications will start in the US. Probably some app will be developed and IT workers will focus on primarily analysis. Note that Palantir CEO Alex Karp majored in philosophy.
As for investing, my suggestion is still Mag 7 or QQQ. It is hard to pick winners at this point.
The good time to invest in AI could last another 2-3 more years [7].
References:
*Google's chip challenge to Nvidia (12-19)
*OpenAI's Code Red, Sacks vs New York Times, New Poverty Line? (12-6)
*Favorable AI market still has 2-3 years, says Deepwater's Gene Munster (12-23)
*LLMs will be stressed by enterprise systems, says Wedbush's Sherlund (12-20)
Dan Ives stays bullish on Microsoft amid AI and mega-cap tech worries (12-23)
Other References:
Scott Bessent: Fixing the Fed, Tariffs for National Security, Solving Affordability in 2026 (12-23)
*'SIGNIFICANT LABOR PROBLEM': Expert reveals US concerns (12-20) Rick Rieder
*CEO warns: AI will reshape the workforce faster than anyone expected (12-15)
*WSJ 12-19 This Is What the World’s Smartest Minds Really Think About AI
Comments