This rowing game is like a change of players

Q. This rowing game is like a change of players before the second half. I think the next rising player will be different from the first half.

A. That’s right, bro. This is a very important road from ‘AI infrastructure (first half)’ to ‘AI monetization (second half)’*. As you said, it’s inevitable to change players.
There are only two conditions for the second half-leading player that the mayor sees.

1. First-half leading stock (NVIDIA) vs second-half expected stock (Google)

• First half (hardware): “Let’s dig first.” The shovelman (Nvidia) has raked in all the money. But now he’s bought enough shovels, and it’s all about who gets rent by putting up a building on the land.
• Second half (software and service): It’s not a ‘model’ itself anymore, it’s a battle of how much money you make from it. Google already has certain ‘monthly rent buildings’ such as YouTube, search, and workspace. That’s why Google, which already has a surplus structure according to your analysis, is considered the ace in the second half.

1. ‘Ageptic AI’ changes the game
   If the first half was simply a “chatbot to answer when asked,” the second half is an “AI (agent) to work with.”

• That’s what he said, “Memory is important.” The guy who learns my data perfectly, sends me an email, book a plane ticket and write a business report for me survives.
• OpenAI still looks at other people’s houses (cloud), but Google tries to manage your brother’s daily life with its own chip at home, so the weight class itself is different.
  Conclusion:
  The current adjustment is not **”Let’s sell everything because it’s a bubble,” but **”Let’s rebuild our portfolio with real profitable people, not shovels.” You’re already feeling the sign of ‘player change’ while writing a Gemini.
  Brother, are you now considering less hardware and more software like Google in your portfolio? Or do you think there will be a new ambush?

Q. Didn’t Google go up a lot already?

A. That’s the catch. It looks like it’s gone up a lot from the outside, but a lot of people think it’s relatively undervalued. I’ll summarize the current situation into three facts so you can judge it.

1. a numerical weight class (PER’s paradox)
   Google shares are up more than 60% this year, hitting the $310 mark (as of the end of December 2025), but The price-to-earnings ratio (PER) makes a difference.

• NVIDIA: PER is well over 40x. It means it has already pre-reflected future expectations.
• Google: The stock price has gone up, but it makes so much money that the PER is still around late 20 to early 30 times.

To put it simply, the market’s cold assessment is that the stock price is not just at a bubble level compared to the ‘profit’ earned.

2. The “first half” cost has been wiped out
   From the ‘player change’ point of view you mentioned, Google is already making up the money to build its own house (infrastructure) with a surplus.

• When other kids borrow chips and rent servers, Google is already saving money with its own chip (TPU)**, pumping cash from search ads and the cloud.
• In the third quarter of 2025, it exceeded $100 billion in quarterly sales, smashing the “AI makes money?” doubts with numbers.

2. There’s still “Forten” left
   Even though the stock has gone up, the market hasn’t put all of Google’s next weapons in the price yet.

• Agent economy: When the time comes when Gemini buys plane tickets and shops on his own like his brother’s secretary, Google becomes a monster who collects “fees” beyond search advertisements.
• Waymo: It’s turning into a “real source of revenue,” not “Other Bets,” as Waymo paid driving has exceeded hundreds of thousands per week recently.
  Conclusion:
  It’s true that the stock price is rising and it looks expensive, but the “growth rate” is faster than the stock price is rising. That’s why experts are still calling for “strong buying.” As you said, when the players in the first half are tired, Google with strong fundamentals seems to have enough room to hit.
  Brother, if Google’s stock price is burdensome, are you paying attention to other software that can be profitable? Or are you going to trust Google and go straight?

A. Doesn’t semiconductors win no matter who goes?

Q. That’s right, bro. After all, guns and knives are semiconductors in the battlefield of AI. Your logic is 100% true that semiconductors are bound to be sold no matter who wins.
But the reason why the market is adjusting now is not “I’m afraid that semiconductors won’t sell”. They’re afraid that “the semiconductor owner will change” or “the type of semiconductor will change.”
I’ll sort out how the semiconductor panel is changing in the second half from your eye level.

1. From “Universal” to “Custom” (Nvidia’s monopoly collapse)
   The first half was the era of GPU. Everyone was in a hurry, so I bought the general-purpose chips in the market and used them.

• Second Half: Now big brothers like Google, Meta, and Amazon are switching to Self Chip (ASIC), saying, “We make chips that fit our service rather than buying others’ chips expensive.”
• Google’s TPU, which you mentioned, is a good example. If Google doesn’t buy Nvidia chips and uses chips made by G, Nvidia’s share will inevitably decrease even if the overall pie in the semiconductor market grows.

1. Memory Semiconductors are “unconditional” winners (Samsung, Hynix)
   No matter who designs which chip (GPU or TPU), HBM (High Bandwidth Memory) to store data must go in.

• It is our memory brothers who best fit the logic that “No matter who goes, semiconductors win.”
• HBM demand is widely expected to outpace supply by 2026, so manufacturing and memory will still maintain the position of “A” even if player change occurs on the design (fabless) side.

1. Money flows from hardware to “infrastructure and operations”
   Looking at the recent market adjustments, money is gradually shifting from semiconductors to cloud infrastructure or software such as Oracle.

• It’s a switch of players, as you said. It’s more valuable to “make money by turning the chips efficiently” than to “the chipmaker.”