Micron Just Hit $1 Trillion. I Bought During the Pullback. Here Is What I'm Doing Next.

Hinzugefügt: 2026-06-01

[00:00:00]A few weeks ago, Micron dropped 14% in 5 days. People were nervous and I get it.

[00:00:06]The stock had already run from $92 to over $700 in less than a year before that dip. So, a pullback like that felt like it could be a start of something worse. I remember staring at the chart thinking I might be catching a falling knife. I bought twice anyway. 1 year ago, the stock was $92. Today, it's around $940. And yes, that's not a typo.

[00:00:32]On May 26th, Micron crossed $1 trillion in market cap for the first time. And I want to quickly explain what that actually means because people hear that $1 trillion number and picture some round number on a stock price. And it does not work that way. Market cap is just the stock price multiplied by all the shares outstanding. Micron has roughly 1.08 billion outstanding shares, which means the $1 trillion threshold was crossed when the stock was trading around $929.

[00:01:04]Today at 9:40, it is sitting just above that milestone. The price action you are seeing right now is not purely driven by fundamentals. It is actually a very interesting mechanic story that I will explain later. So, let's talk about how did Micron reach the $1 trillion number?

[00:01:22]So, a few things led to that. A UBS analyst raised his price target from $535 to over $600. The new price target did not cause the stock price to jump. It was the argument behind that target price. UBS basically said the market needs to stop treating Micron like a commodity memory company. Long-term supply agreements with hyperscalers are locking in pricing years out. The demand for high-bandwidth memory is not acting like a cycle anymore. Micron's entire 2026 high bandwidth memory supply, including next generation HBM4, is already sold out. CEO Sanjay Mehrotra confirmed that Micron can only fulfill about 50 to 65% of what its key customers actually need right now. Their best customers cannot even get everything that they are asking for.

[00:02:18]This gives Micron the pricing power. SK Hynix also crossed $1 trillion joining Samsung, which got there early May. All three major HBM producers are now trillion-dollar companies. The whole sector is being re-rated. Revenue last quarter was up 196% year over year, beating Micron's own guidance by more than $5 billion. The forward PE is roughly nine times. The semiconductor industry median is around 36 times. So, the market is not accidentally pricing Micron this low. Memory companies have a long history of explosive earnings followed by brutal collapses. With additional capacity, prices usually normalize. What looked permanent turns out to be a temporary thing. The market remembers that and is hedging against it happening again. And this is not the first time markets have overbought permanent demand themes in the past.

[00:03:17]Look at telecom, fiber, solar, NAND. The pattern of building too much around a real trend is not new. But UBS is making the argument that this time the dynamic is different with the AI boom. HBM is not selling into consumer devices that go in and out of fashion quickly. It is selling into AI data centers, which are under long-term contracts with hyperscalers who have publicly committed to spending hundreds of billions on infrastructure. We have never seen this kind of demand before. So, the nine times forward PE is either a warning or it's an opportunity depending entirely on which of those two views turns out to be correct. I personally lean towards what UBS is saying, but you should also know that the average analyst price target on Micron is about $758. The stock is now at 940. It has already blown past what the average Wall Street analyst thinks it is worth. But, they have all been wrong about it all year.

[00:04:21]So, is Micron a buy right now? And I will also talk about what I am doing.

[00:04:26]Well, this is the question I get a lot, so I want to share my perspective on it.

[00:04:30]Some analysts have a $2,000 price target on Micron, but even if that is right, it does not mean buying at $940 today is a smart move. The stock has already moved 860% in the past year. A lot of that easy money has already been made. Part of this move is not just investors buying because they believe in the business.

[00:04:54]Goldman Sachs actually flagged this as a historic gamma squeeze. Here is what that means in simple terms. When traders buy call options in massive volumes, the dealers who sell those options are forced to buy the actual stock to protect themselves. And as the stock rises, they have to keep buying more.

[00:05:14]That buying pushes it higher, which triggers more call buying, which forces more hedging. So, it feeds on itself. In the session that led up 19% single-day jump, over 53,000 contracts traded on the $1,000 call alone. That is the feedback loop made visible. And when it stops, it can reverse just as fast. Micron also guided to roughly $19 per share in earnings just for next quarter alone. If Micron can sustain something close to that earning space over the next several quarters, the stock is not nearly as expensive as the headline move suggests.

[00:05:56]But memory earnings are volatile and one strong quarter is not a permanent thing.

[00:06:02]So the real bear case is that the supply will eventually catch up enough to start compressing margins even while demand stays strong. Memory companies do not need falling revenue to start disappointing investors. They just need pricing power to weaken. And that is how memory cycles have historically ended.

[00:06:22]Profitability quietly normalizes while everyone is still excited about the growth story. The price where I am more comfortable is somewhere between $760 and $800. At that level, you are paying eight times forward earnings, which is the multiple the market has historically applied to memory companies. I don't want to be paying up for a stock in the middle of a momentum surge after an 860% run. I want to buy when the market is nervous, not when it is excited. As you all know, I bought twice during the last pullback and I feel good about those purchases. I am not adding at 940. I will be watching the June 24th earnings report, specifically whether demand visibility extends into 2027 and if the story keeps getting stronger and the stock pulls back, that is when I get more interested. I am not selling at 940. I feel more interested in protecting what I have than aggressively trying to add to it. I also want to address the question I keep getting about the DRAM ETF. I completely understand the appeal. It is up over 90% since April, now over $12 billion in assets. It feels like a diversified way to play the whole HBM story without betting on one company. But, here is how I actually think about it. The fund has 15 holdings, but the top three make up over 73% of the portfolio. Micron is sitting at around 28% SK Hynix at 28 and Samsung at about 18%. You are still making an extremely concentrated bet on those three companies tied to the exact same AI infrastructure cycle. That is not really diversification.

[00:08:10]Then there is the cost. DRUM charges 0.65% annually in fees. That is roughly three to four times what you would pay for a typical broad market ETF. So, on a $10,000 position, that is about $65 per year in fees regardless of how the fund performs. The tax piece is also worth understanding before you buy. The fund uses total return swaps and is actively managed, which creates a more complex tax situation than a typical passive ETF. Distributions are paid annually in December. The structure makes tax efficiency lower than a standard ETF and that is something you should factor in if you're holding this in a regular brokerage account. There is also a geopolitical angle that people overlook.

[00:08:57]Over 45% of the holdings are South Korean equities including SK Hynix and Samsung. Roundhill's own prospectus explicitly names North Korea tension as a risk factor. Half of what you own stocks trading at 3:00 a.m. Eastern.

[00:09:15]Overnight news out of Korea creates gaps you cannot react to. That structural disadvantage does not go away. And since launch, Micron has actually outperformed the ETF anyway. So, for me owning it directly just makes more sense without the added fees and the tax complexity and also the overnight gap risk.

[00:09:38]To be fair, the strongest case for DRAM is that it protects you if you are wrong about which HBM company wins. If Samsung catches up or SK Hynix pulls further ahead, the ETF captures that upside, too.

[00:09:52]And that is a legitimate reason to choose it. I just personally think Micron is the better bet right now.

[00:09:58]Before I wrap up this video, there are several risks that you should know about. Samsung is the biggest one. They have been struggling with manufacturing problems in HBM, but they are not going to stay down forever. If they fix those issues, the supply shortage driving Micron's pricing power goes away faster than anyone expects. And if you look beyond Samsung, the broader risk is that the semiconductor industry is very good at solving shortages. New capacity comes online, the yield improves. What feels like a permanent bottleneck can normalize faster than the market anticipates. And then there is the cycle question. Memory has always been cyclical. Every time someone said this cycle is different, it eventually normalized. Maybe AI changes that permanently. Maybe it doesn't. That uncertainty is real. The June 24th earnings report is also a risk. Wall Street is expecting roughly $24 billion in revenue next quarter. That is an extraordinarily high bar. If Micron just meets it without blowing past it, the stock could sell off even on a technically good report. In this scenario, high expectations are dangerous. A few weeks ago, I told you I was buying during the pullback. The view was that AI demand was structural and the market was under pricing the transition from commodity to infrastructure. That view has now crossed $1 trillion, and I'm not chasing it here, but I'm not selling either. Micron is the only major US based pure play memory company directly at the center of the HPM bottleneck that is defining the AI infrastructure buildout. And whether that bottleneck lasts two or more years or five is honestly something I am still working through.