What to Watch as Micron Reports Earnings Hitting on All Cylinders

SPEAKER_1 (0:00)

All right, let's turn our attention straight to our tech spotlight and preview Micron's earnings. That report is due out after the close today. Joining me for this conversation, Stephen Sopko, Practice Lead Semiconductor and Deep Tech HyperFrame Research. Stephen Sopko, great to see you. Some of your expectations for Micron later today.

Stephen Sopko (0:24)

I think the thing we have to realize is that Micron is hitting on all cylinders. We don't really expect any surprises, but there's a lot of things we're looking for, particularly in the transcript, that are going to give us a hint of what 2027 looks like, particularly for that high bandwidth memory.

High bandwidth memory is the memory that AI really relies on. Only three companies make it at scales. You've got Micron, SK Hynix and Samsung. Samsung actually said something very interesting this week, that they see the percentage of wafers, the materials that memory is made from, that are going to be allocated to that high bandwidth memory is going to jump from about 20 percent of the global supply up to 30 percent. That is a finite supply of wafers and you have to decide, do I make high bandwidth AI memory out of it? Do I make high quality server memory that AI relies on? Or do I make consumer devices and all of the other things with memory? A lot of what we look for with Micron, particularly in the earnings, is hints on where they see that market going, and whether they're going to be able to meet the demand in 2027, because that has been something that's been a big question all the way around.

SPEAKER_1 (1:41)

Look, analysts are expecting Micron to report fourth quarter revenue of over 50 billion, that they've seen the trajectory of beating on a regular basis. I think they beat revenue 12 straight quarters in a row.

So, it's not a question of whether or not it's a well-positioned company, but to your point, the high bandwidth memory and DRAM orders going forward will be front and center, and also how high bandwidth memory is growing more, as you noted, more to that 30 percent. So, do you expect them to really just give it to us and say record orders are coming in? Are we going to hear that kind of language, particularly when you have Korea also adding so much in this space too, which I want to get to?

Stephen Sopko (2:24)

Absolutely. I think the thing we have to realize is Micron is historically a very conservative company, and so they tend to not go for fireworks in their marketing or their transcripts. So, what I expect is just really a clear read on where the business is right now, and I'm looking for an indication of, yes, you're sold out of HBM through 26 We've known that for a while. You've indicated that possibly there's a similar situation all the way through 27

When you combine this memory shortage, if you will, with the demand for CPUs, the orchestration chips that make AI work, and you've heard Intel talking about not being able to meet all of their orders, similar signals coming out of AMD and even Nvidia. So I think that there is so much demand across all layers of this market, that what I'm really looking for from Micron is some indication as to where that's going next.

SPEAKER_1 (3:30)

You did mention going forward, right? You mentioned these memory shortages in your notes, and that's something that has certainly caused some volatility in the market, that supply-demand chain and that really needs to go smoothly. What about the outlook for 27-28 on the supply chain, the demand and the supply meeting, rather than having shortages? Because then you have the pricing power or the pricing gate.

We heard, for example, Apple say they have to raise because the prices of the iPhones, because memory chips cost more. Will this even out? You want to have the demand, but you want it to even out.

Stephen Sopko (4:10)

That's true. I think the thing we have to realize is capacity takes years to grow. It's funny because I'm at a CoreWeave event this week and I'm listening to enterprise customers who are radically experimenting with AI. They are doing some very clever things, all within guardrails, all safe, but you can see how it's going to fundamentally transform their businesses. That's such a clear demand signal for AI going forward that I think when a company like Micron has a limited or a fixed number of wafers they can build from, they've got to decide, do I use those wafers to build that AI memory, to build the server memory, or to build consumer memory? That's the Apple conversation. Not necessarily Micron, but any memory maker.