I want to ask everyone a question.

Assuming that Micron (MU) will likely oscillate between 800 and 1200 from July to October this year, but has a chance to see 2000 by Q1 2027.

What would you do now?

Would you cut losses and exit due to short-term adjustments, or gradually buy in batches during the pullback?

Yesterday, I gave you a set of data: There is one set of data worth remembering: In U.S. stock market history, whenever the S&P 500 rose more than 10% in Q2, it only fell in the second half once (1975), and on average, the second half typically rises another 12% or so, and Q4 has never fallen in history after such a situation. Then, after the midterm elections, uncertainty dissipates, and Q4 usually performs the best.

I’ll post it again—everyone should at least stick it out until the end of this year.

My own view hasn’t changed.

I believe the AI mega-cycle is far from over. If someone thinks we’ve already reached the craziest point, I don’t agree.

I would compare the current period to the first half of 1998. Coincidentally, 1998 was also a midterm election year, and Powell’s philosophy is very similar to Greenspan’s (no rate hikes this year; Powell is hawkish on the outside but dovish on the inside—I guess either staying put or cutting rates). Back then, the internet boom had already started, but the phase of true mass frenzy, where people would buy anything and everyone felt "there’s no bubble," was from late 1998 to early 2000.

The top won’t appear when people are afraid; it only appears when the market is extremely euphoric and irrational.

Looking at the current situation, the most common discussions every day are still: "Is AI over?" "Will capital spending be cut?" "Has semiconductors peaked?" Or when everyone is silent and there’s dead stillness—that’s when you must pay attention to buying the dip. Just like when everyone despises BTC and thinks it will go to 30,000 or 40,000, or calls it a scam—you need to think contrarian and force yourself to pay attention. It’s the same with AI now.

I find it hard to associate this sentiment with a bull market top.

At least so far, what I see is earnings and demand driving industry development, not purely sentiment pushing up valuations.

Also, the market has been worried about cloud vendors cutting AI capital spending. But don’t forget, the major cloud vendors have announced combined AI capital spending of about $725 billion this year. This is the annual plan set at the beginning of the year. I think the likelihood of a significant reduction in the short term is low, and the earnings season for major companies at the end of July will prove that.

If AI deployment continues to advance over the next few years, there is still room for further growth in industry capital spending. What we really need to worry about is if companies pour in huge amounts over the long term but the return on investment (ROI) fails to keep up, and cash flow remains under pressure—only then could capital spending enter a contraction cycle. But in my view, that’s something to keep an eye on over the next few years, and we won’t need to worry about it until at least 2028. It’s not something that has already happened just because the market is down today.

At the same time, AI inference, agents, and enterprise deployments are increasing, token consumption continues to grow, and the underlying computing power demand is still expanding.

So my view hasn’t changed.

If the long-term logic hasn’t changed, short-term fluctuations are just part of the market.

A bull market doesn’t end in pessimism.

Remember that sentence.