AI & Software (2-8-2026)

Jia Han
3 minutes ago
6 min read

This is likely an inflection point for AI and IT. Several things led me to this conclusion.

Brad Gerstner had an interview on CNBC [1]. ([2-4] were its clips; they are included in case [1] does not work.)
Some interviews from last Davos (World Economic Forum): Jensen Huang, Alex Karp, Sundar Pichai (Google), Satya Nadella (MSFT), Meta AI Chief, etc. [5-7].
Recent earnings reports.

At the start of this year, Brad Gerstner said that falls in 90% software stocks were justified. The last few weeks have proved that he was correct. CRM, Adobe, Workday, and many others continued falling. The main reason is paradigm change in the IT industry.

It will be helpful if we recount which one, hardware or software, is more important in the computing industry. In the 1960s, IBM was the leader in computing and it produced both hardware and software. In the early 1980s, Steve Jobs invented the Apple personal computer. IBM was late to the PC market and thus was forced to contract out to others to make personal computers under its name to compete. IBM made the judgment call that hardware makes money while software is just peripherals. The value of software was only to sell hardware. This was true for mainframe computers. Few anticipated that computers, PC in particular, would find so many applications. Microsoft’s software has become ubiquitous. Once an application is developed and tested, it costs virtually nothing to duplicate. Microsoft becomes more valuable than IBM now.

However, writing software is still very demanding. If you have worked as a computer programmer once in your career, you would know that debugging can be both frustrating and time-consuming. The whole process may be viewed from top-down, first the architecture, then algorithm analysis, coding, debugging, and finally testing. At the machine level, all programs consist of digits of 0s and 1s. Software development is expensive because a human does not think flawlessly like a machine. A human being makes mistakes but digital machines do not. An application software needs extensive human effort to code and debug.

This has changed with the progress of AI. AI can automate a vast amount of coding and debugging. In a recent interview, Mark Zuckerberg said that at Meta, one person can do more than ten workers do. Think about the old approach, which used to involve many workers. Now much work has been automated.

There are several things unclear, and future development may change significantly. First, Jensen said that AI has five layers. He might be right, but this could depend on future developments. Second, there is ferocious war among the infrastructure built-ups: Google, OpenAI, Meta, Microsoft, Anthropic, Amazon, xAI. As Jensen said, Meta is the first that successfully applied AI and increased productivity. Recent results Anthropic tools show powerful automatic programming. Open AI is ambitious and gets a lot of funding. But its cashflow (debt financial) is a bit of concern. Google has very good technical expertise and its CEO

the future of each company, and its stock performance, depends on its CEO’s vision. Take Microsoft as an example. The previous CEO of Microsoft, Steve Ballmer, missed two inflection points: mobile computing and cloud computing. It was the current CEO Satya Nadella, who made adjustments so that Microsoft would catch up. However, I watched his interview on Davos. I am not sure that he has the vision to guide this transition.

How a company performs in this computing paradigm shift depends on several things. One is whether its top executives have a good vision. Another is the products and customers. So far, Meta is leading. But its customers and products are relatively easy to change the computing paradigm. Microsoft is harder because many of its products are legacy types embedded in its operating system. New company Anthropic is developing new, impressive applications and tools. Companies are reluctant to change software. However, if new ones are much cheaper and easy to upgrade, many may consider. All companies will face big challenges, Microsoft more so. Other big tech companies are OK. Google is in a good position because it has talent, and its CEO Sundar Pichai has a good vision. [8] has more on software debate.

Microsoft is unclear. Whether its CEO has a good vision can be very important. Its previous CEO Steve Ballmer missed two new trends, mobile computing and cloud computing, Microsoft underperformed during Ballmer’s tenure. Its current CEO Satya Nadella corrected these mistakes and renewed Microsoft. Microsoft made an investment in OpenAI and publicized ChatGPT, many were excited and thought Microsoft would lead the AI revolution. It is not that simple. LLM was open source and open architecture, thus Microsoft had only a small advantage. After watching Nadella’s interview at Davos, I was not impressed by his vision.

Investment

First, a safe way to participate in the AI revolution is still QQQ. My writing above is useful only if you already have a good strategy. Then you may adjust accordingly. One important note on semis. If you have basic knowledge, you would know that semi stocks are cyclic (check past books or the Internet to see the meaning of semis cycles). This time it could be different. In the past, you had one CPU. Now AI computing needs great computing power and needs many chips. The recent memory chip stocks rapid rise shows this. Memory chips, however, are hard to predict because they are commodities. In Update 1-10-2026 (Trade, AI) I mentioned chip equipment stocks AMAT, KLAC, ASML, LRCX, chip stocks: AVGO, AMD, NVDA. Cyber security is another area: CRWD, PANW. Chip equipment stocks are safe to invest since most still underestimate chip demands.

Jobs

My guess is that high level (architect, analyst) jobs will still be in demand. I pay attention to Palantir Alex Karp’s interview because Palantir is the only company that applies its AI expertise to many disciplines. But his answers were not clear. My guess that high level jobs are still needed seem correct. If you work downstream, for example, financial analysis, you probably will not know until some tools are in commercial use.

(from Jim Cramer email 2-5. Concerning the US job market)

2. More evidence of a shaky job market: Outplacement firm Challenger, Gray & Christmas said its monthly look at planned layoffs at American companies came in at its highest January total since the global financial crisis. Hiring intentions reached their lowest levels since the same period. The data this morning followed yesterday's tepid private-sector job growth report from ADP. The government's January employment report, which was supposed to be out tomorrow morning, was delayed due to this week's brief government shutdown.

References:

Philosophy, Sciences, Religion

AI & Software (2-8-2026)

Recent Posts

Comments