Micron Stock: The Most Undervalued AI Play? | MU Price Prediction 2026

追加: 2026-05-16

[00:00:00]Hello everybody. Today we're going to do Micron Technology, whether it's a buy, do we still have room to run in this memory supercycle, and what are some things that we should look out for as the memory supercycle is running hot.

[00:00:14]Micron is at the center of this memory supercycle in the AI.

[00:00:19]It has a quarter two reports are at $23.9 billion.

[00:00:25]58% margin, 11 4 PE, and $720 billion market cap.

[00:00:31]Ticker symbol MU.

[00:00:32]So, Micron is the only US-based manufacturer of DRAM and NAND flash.

[00:00:38]It's a critical node in the global semiconductor supply chain and a national security asset. It's considered to have several core products, specifically the high bandwidth memory and AI accelerator backbone. DRAM, NAND flash, LPDDR, and CXL memory. And in fact, it's basically it has a monopoly on majority of these products here in the US.

[00:01:00]Business units has got the cloud memory, core data center, mobile and client, and automotive embedded. It was founded in 1978 Boise, Idaho with 48,000 globally.

[00:01:11]It has two competitors, Samsung and SK Hynix.

[00:01:15]Hynix.

[00:01:17]It's actually HBM. Both of them are HBM competitors, but they're located in Korea.

[00:01:22]The US Chips Act was a huge tailwind for Sam for Micron. Specifically, it was awarded $6 in federal grants, and its HBM status has already sold out for 2026 under contract, and its key customer is Nvidia for both its Blackwell and Vera Rubin platform. So, if we take a look at the ecosystem of where the cap expending is going, a lot of it's going into Micron because of what's happening with um Nvidia's chips.

[00:01:53]Now, financial performance for the last couple of quarters has been record-breaking. In fact, Q2 of this year Micron reported $24 billion in quarterly revenue. It's highest it's ever been since 2024.

[00:02:09]On average, it's had about 8 to 9 to 14 billion dollar in quarterly revenue. But it actually improved quite substantially to 196%.

[00:02:20]For full year, it's increased revenue at 49%.

[00:02:25]And projected for revenue guidance, 33 billion for Q3. And this is just a management estimate. But based off the way the trajectory going is going, it's going parabolic. Gross margins are expanding at 68. So as you can see from this chart, basically this year was a banner year for Micron as the demand for memory has just skyrocketed due to obviously the AI data center rollouts and the usage for AI.

[00:02:53]Now, HBM was considered to be a commodity in the past. And this is where things start to shift a little bit because of what's happening right now with the growth rates. In fact, if you take a look at the HBM market share, primarily SK Hynix has 62% of the market share, which obviously would be the top dog. In second case will be Micron and third will be Samsung.

[00:03:18]However, this actually is a little skewed because these companies are located in Korea.

[00:03:24]The HBM addressable market for 2026 is $50 However, it's estimated that's going to double to 100. So that pie is just going to get a lot larger.

[00:03:36]They're already sold out for 2026. All HBM 4 capacity is out. They are under binding contract. So pretty much they can't get out of those. So the revenue is stuff is already locked in for this year.

[00:03:49]And there is revenue visibility for going in for the next 2 years.

[00:03:54]Now, the server premium is really where things start getting interesting because the AI servers need six of the DRAMs and eight of the NANDs versus the traditional servers. So, in other words, they're going to need more chips for each of the servers compared to historically. Number three, the PowerEdge.

[00:04:10]30% of the PowerEdge, they use less 30% less power than the rivals. So, in other words, hyperscalers are going to want to buy those relative to cost as they start ramping up the memory chips.

[00:04:23]And finally, the most important one, the multi-year deals. Um Micron has already secured 3 to 5-year supply that has replaced quarterly pricing, and this should dampen the cyclicality. In the past, memory has been a cyclical industry, meaning that they basically have their ups and downs, but because of these multi-year deals, um it's going to pretty much have a higher revenue visibility.

[00:04:51]Now, Micron is the only US player in town, which will create a massive monopoly in the US market specifically.

[00:04:58]So, we talked about the three that we talked about earlier, its market share is at 21%, but it's growing very, very quickly. Samsung is recovering, and SK is the leader, but if things start to change really quickly or re-rate, what's going to happen is Micron is going to eventually potentially be number one.

[00:05:19]Now, power efficiency is better between the two. Number three, the HBM node, it has one gamma in HBM4, while the other two are still kind of catching up.

[00:05:30]So, HBM4 is going to be your next evolution of the memory chips, which if Micron starts to rebuild, they're going to start to actually have a larger case in it.

[00:05:40]The bigger part is going to be the Nvidia partnership, specifically with Blackwell and Vera Rubin.

[00:05:46]SK and Samsung have a supplier, but they're still qualifying for it. So, my take on this is the Nvidia partnership is actually a key part in the revenue stream down world.

[00:05:57]The US policy tailwind has a 6.4 billion dollars chips act, while the other two don't.

[00:06:04]And finally, they're all sold out, while Samsung still qualifying.

[00:06:08]Now, the manufacturing plants is in Idaho, and New York, and Virginia, plus in Japan, while the other two are only in Korea. This provides diversity on the manufacturing side, and potentially could actually have more markets, access to more markets, larger markets in the future. And finally, for the valuation.

[00:06:28]This is where a lot of people kind of confuse the valuation from the past to the current. In the past, these industries usually have low valuations at the peak of their cycle.

[00:06:38]In other words, earnings are growing very, very quickly, and that usually signals an end or a top. So, here is right now a little word of caution. If we rewrite those back to the way they were in the past, that is considered to be topping signals. However, I believe that the market's still using the old model of of valuing up these these companies. So, 11 is the cheapest.

[00:07:03]The other two have a higher multiple.

[00:07:06]Now, I believe that they are superiorly mispriced at this point, even at 11 times earnings for a couple of reasons.

[00:07:12]Number one is that majority of the analysts have a buy target for Micron at 742 to 525, which I believe we are close, if not lower than where we're supposed to be at. So, the analyst has to still rewrite these a lot higher. Number one is they have an 11x forward PE, which is the consensus, which is I believe it's still on the low side, assuming it is a growth company, which I believe it is.

[00:07:38]Number three, uh Um, bull case is running at 750 to 1,500.

[00:07:43]So, the there's a lag period right now with the re-rates because of the way that they're doing their models.

[00:07:50]The PEG ratio is 0.07, which is their earnings growth, which is growing double digits. I believe that that is still undervalued. Their revenue for, um, 2026 is 108 billion.

[00:08:02]And they have already updated their forecast. So, if we take a look at the earnings inflection of what's happening with the trajectory.

[00:08:09]For 2027, they were already estimated to be at about 30 billion at 30 at 30 versus at the 19 level.

[00:08:17]So, in my opinion, I think this is going to be something very interesting because if you take a look at where we are at right now, it's trading of as if it was a dying cyclical.

[00:08:29]This has historically been true if you go back to the dot-com era when memory chips were key components in computer systems like desktops or regular server racks.

[00:08:41]But, because we're dealing with AI server racks, which use much more memory and more compute, that has changed the entire, um, calculation.

[00:08:51]It's generating 190% revenue growth and with 58% net margins.

[00:08:56]These margins are extremely high for a hardware company.

[00:09:00]And a 5.9 exit PE is already priced in, which means if the market does re-rate, Micron can even 20x or half of Nvidia's multiple. So, the stock basically doubles from here in layman's terms.

[00:09:13]Now, in terms of upside, obviously, we're look like we're kind of capped out here because a double is not necessarily something that people are going to be looking at. And Micron was probably better to be bought a couple of months ago or couple years ago, but nobody would have known that the servers would have been using that much RAM. So, the question here is is this still a value?

[00:09:34]The answer is if you believe that the AI super cycle is still here and then the demand for the memory chips are going to start to move very very quickly, then I think we're still early in this entire cycle.

[00:09:47]Now, what can go wrong? So, let's talk about the risk registers here real quick. So, the memory cyclicality is a true concern from the bears. The first thing is memory markets have always boom boom bust cycles and it's because hyperscalers capex starts to grow and if they over order, they're going to stop ordering which in turn it's going to be a problem. So, if we take a look at the timeline, 2027 could be the peak which is next year and these multi-year contracts all capacities did they just basically reduce the AI risk but they don't eliminate it. So, I'll give you a scenario. Let's assume you buy it now, you wait till next year, there could be a chance when all the hype capex spenders have already had the memory so they don't need to upgrade their systems.

[00:10:30]Quite possible. Number two, China.

[00:10:32]20% of the revenue is exposed to China and the products are banned from Chinese infrastructure which includes the new tariffs and the supply chain. So, Micron is actively lobbying the government to allow those curves to be out there but this is because of the problems that we're having right now with the current regulations. So, these two for the most part are like the biggest biggest key parts of this whole entire narrative. Number two, for execution risk. So, if HBM 4 is going to be very key with Nvidia and and it's going to allow Samsung to really catch up if they can't roll this out.

[00:11:07]Specifically 2026 and 27 because of the fact that they are still ahead of the game now but the execution is going to be key. Any sort of slow down in that execution, the stock falls.

[00:11:19]Capex burn. So, 6.4 billions in capex, um they usually obviously manufacture their chips. If the demand softens for the fabs, their free cash flow will start to compress and they already have already $200 billion of domestic fat commitment for the long term. Now, they do have $16 billion in cash, and the chip grant offers them a portion of the bill cost. So, this is kind of like a medium one, but again, they're heavy cap expenders because they have to revamp the machines. Finally, for the low compression, so at 11x forward compressions, it's basically it could go back to a 4 5 7x is possible if they lose structure of that. So, that basically means you're looking at about getting cut in half. So, you're looking at a double or a cut in half. So, it's kind of like sitting sits right in the middle. Now, Samsung. Samsung's expected to make a comeback, but this is actually very low. It's a financial um customer, um but again, it's the power efficiency that kind of separates uh Micron from the rest. So, from all of these six, I would say the likelihood of a re-rate to the downside would memory be the memory cyclicality, which at this point right now, there's no evidence of that. And the China deal political one, any sort of positive news can remove this. All of this other stuff, I think it's a medium-to-low risk.

[00:12:35]Okay, key events that can move. So, catalyst what we're looking for for this year and to next year, as we start getting more visibility on cap expending, has to do with the JP Morgan conference on May 20th, where uh Sanjay is going to pretty much talk about the HBM 4 volume and the guidance going in.

[00:12:54]On June 4th, they're going to have their quarterly earnings. The consensus is 33 to 40 billions. Uh most likely they will beat because of the HBM capacity. The question is the guidance going into the following year, and actually to the tail end of Q4. So, how does that look, and how much demand they have in their backlog? The hyperscaler earnings are on July the 26th. They're going to be reporting about CapEx. This is actually a leading indicator. So far, we haven't had any slowdown in CapEx. In fact, it's been continually going up. This is actually um a very important indicator for Samsung. If they start to say they're reducing CapEx spending, then we got a problem.

[00:13:31]The HBM 4 ramp is going to be a later this year. Um they're going to be sending it to Nvidia specifically for Vera Rubin's chips. They're expansion catalyst with 30% power efficiency lead is remote and is a core product, so I think that's going to be very important.

[00:13:46]Now, for Samsung Q4, that's going to be HBM 4 competitor. They want to take a look at Samsung actually is able to pick up and um let's just say catch up to the manufacturing for the HBM 4.

[00:13:59]The last year will be a buyback announcement. They have about $16 billion in cash, and many analysts expect a repurchase of the shares by the year end. So, I think that's going to be a good floor for the stock assuming it doesn't run ahead of that. And finally, the cycle peak. So, if we take a look at the cycle, there's going to be new capacity comes in online this year.

[00:14:18]The supply demand could shift as more memory comes onto the market, and that is the core bear case scenario if we start getting a lot a lot of supply glut. And finally, 2028 2 years out, the TAM value for the HBM market is going to double. So, if analyst projections are correct, which I believe at that point the total addressable market should grow. The Micron um basically estimates that the HBM market will exceed the size the entire size of the 2024 DRAM market. It's structural and it's not cyclical. So, based off these, I believe that there's going to be an opportunity in Teradyne. The stock is definitely overextended at the moment, and there is a lot of in good news already baked in. The question is, at what point do we see a re-rate of the stock so that it can go to the double up from here?

[00:15:07]Okay, so here's my investment verdict, and again, this is not financial advice.

[00:15:11]Please consult a financial advisor for that just because um things can go the other way. All right, so the bull case. Number one is we do have more bull case narratives right now because the market is into this like bull mode with the memory chips. Number one is the number one US memory maker.

[00:15:26]Basically holds a monopoly. So, no competition and it's considered to be a national security asset. So, it's blessed by the government. Number two, sold out for the HBM 4s. 194 90% year-over-year. I think this actually trades like a software company, which is amazing.

[00:15:42]Only 11 4 PE. Downside is pretty much floored at this point assuming things go out the door, which I think it's a pretty reasonable number.

[00:15:51]Nvidia's confirmed as the HBM supplier for Blackwell and Vera Rubin. So, Nvidia is holding a lot of the capex sending about 70% of the capex money. It's going to Nvidia and Nvidia now is a key customer.

[00:16:04]6.4 billion from the CHIPS Act grants a shield and accelerates domestic build.

[00:16:08]So, that's already kind of a little insurance by the government. And we do have HBM 10. So, long-term it's a structural change, not a cyclicality.

[00:16:17]Now, the bear case. We do find that memory is violent.

[00:16:21]China's still there. HBM 4 capex and AI infrastructure demand needs to stay. Our price targets are basically in line with kind of like some of these analysts, but I feel like the analysts are still catching up. Number one bear case 280 to 320, which is about a 50% cut from here.

[00:16:38]Which again, you have to be okay with that. The base case is 480 to 525, still a little bit lower. And the bull case 750 to 15. Now, based off of these news based off of these numbers right now, Micron is trading at and I just have to kind of take a look at those. So, if you take a look at Micron's overall price trajectory, analyst consensus are still at strong buy. Definitely a high conviction one and I would say probably something to add to the portfolio as that. But, please consult a financial advisor. All right, I'll see you guys next time.