Latest / AI Investment Edge: Stocks, Markets & Money / EP 29 | Micron Stock Analysis: Is the AI Memory Boom Priced In?
Transcript
- 0:00SPEAKER_01: You know,
- 0:00SPEAKER_01: when a massive gold rush happens,
- 0:02SPEAKER_01: it's usually the people selling the shovels who make the real money.
- 0:05SPEAKER_00: Right.
- 0:05SPEAKER_00: Yeah.
- 0:05SPEAKER_00: The classic shovel seller theory.
- 0:07SPEAKER_01: Exactly.
- 0:08SPEAKER_01: But what happens when that shovel factory is suddenly making so much cash
- 0:13SPEAKER_01: that they decide to just take those incredible boom time profits
- 0:17SPEAKER_01: and build like five more massive shovel factories?
- 0:21SPEAKER_00: You start to wonder if those profits can actually survive the expansion.
- 0:24SPEAKER_01: Exactly.
- 0:25SPEAKER_01: Do they survive the very expansion they're funding?
- 0:28SPEAKER_01: Well, welcome to AI Investment Edge, episode 29.
- 0:32SPEAKER_01: Today, we are taking a deep dive into Micron technology.
- 0:35SPEAKER_00: And they are,
- 0:36SPEAKER_00: without a doubt,
- 0:37SPEAKER_00: one of the most critical shovel sellers
- 0:38SPEAKER_00: in this entire artificial intelligence gold rush.
- 0:41SPEAKER_01: Yeah, they really are.
- 0:43SPEAKER_01: And today,
- 0:43SPEAKER_01: our core mission is to examine that memory scarcity,
- 0:45SPEAKER_01: break down this massive capital cycle,
- 0:48SPEAKER_01: and really figure out the price you pay as an investor.
- 0:51SPEAKER_00: Because of that tension,
- 0:52SPEAKER_00: that massive demand versus the incredibly expensive process
- 0:55SPEAKER_00: of building new capacity...
- 0:57SPEAKER_00: It's the defining puzzle of the semiconductor industry right now.
- 1:00SPEAKER_01: It really is.
- 1:01SPEAKER_01: And just to give a quick overview for anyone who is an elbow deep in the tech space,
- 1:06SPEAKER_01: Micron basically makes vital AI components.
- 1:09SPEAKER_01: They produce DRAM, NAND storage, and high bandwidth memory.
- 1:14SPEAKER_00: Right.
- 1:14SPEAKER_00: The stuff that makes these AI systems actually run.
- 1:17SPEAKER_01: Right.
- 1:18SPEAKER_01: And because of that tension we just mentioned,
- 1:20SPEAKER_01: I want to state the current published iPulse AI consensus view right off the bat.
- 1:24SPEAKER_00: Let's hear it.
- 1:25SPEAKER_01: So for the one-year horizon ending in the third quarter of 2027, the consensus is a buy.
- 1:31SPEAKER_01: But if you look at the five-year horizon ending in the third quarter of 2031,
- 1:36SPEAKER_01: it drops to neutral.
- 1:37SPEAKER_00: Which is a pretty significant shift.
- 1:39SPEAKER_01: It is.
- 1:40SPEAKER_01: So the immediate combined stance is no action, keep monitoring.
- 1:44SPEAKER_00: And I should clarify the price basis for those ratings immediately
- 1:47SPEAKER_00: so you listening know exactly where this starts.
- 1:50SPEAKER_01: Yeah, that's important.
- 1:51SPEAKER_00: These current ratings are using the October 23 market close,
- 1:55SPEAKER_00: which was shown at about $878,
- 1:57SPEAKER_00: and they use the remaining time to those fixed targets you mentioned.
- 2:02SPEAKER_01: So it's an update, basically.
- 2:04SPEAKER_00: Right.
- 2:04SPEAKER_00: It updates the price and time reading of the existing September deep research.
- 2:08SPEAKER_00: It is not a brand new from scratch deep analysis run.
- 2:11SPEAKER_01: Okay, got it.
- 2:13SPEAKER_01: So let's jump into the actual financial reality from that September research,
- 2:17SPEAKER_01: because I was looking at their September 30 results and the October 9 annual filing,
- 2:21SPEAKER_01: and I mean, the numbers are just staggering.
- 2:24SPEAKER_00: Oh, absolutely.
- 2:25SPEAKER_00: The scale of the AI boom on their financials is wild.
- 2:28SPEAKER_01: Yeah, like annual sales climb from roughly $37 billion to $133 billion.
- 2:35SPEAKER_00: Incredible growth.
- 2:35SPEAKER_01: And net income is sitting around $85 billion.
- 2:39SPEAKER_01: But the analytical key here,
- 2:41SPEAKER_01: and this is fascinating,
- 2:43SPEAKER_01: is that memory pricing rose much, much faster than their actual shipment volumes.
- 2:48SPEAKER_00: Yes, which means pricing durability is the absolute center of the investment case.
- 2:54SPEAKER_01: Right, because they didn't just ship three times as many chips,
- 2:57SPEAKER_01: they just charged way more for them.
- 2:59SPEAKER_00: Exactly.
- 3:00SPEAKER_00: And you can see that clearly in their profitability.
- 3:03SPEAKER_00: Their gap gross margin jumped from about 40% all the way to 81%.
- 3:08SPEAKER_00: Wow.
- 3:08SPEAKER_01: 81%.
- 3:10SPEAKER_01: Yeah,
- 3:10SPEAKER_00: it's massive.
- 3:12SPEAKER_00: And that clarity on margins is driven by a few things.
- 3:15SPEAKER_00: The pricing,
- 3:16SPEAKER_00: obviously, but also their product mix and just really solid manufacturing execution.
- 3:21SPEAKER_01: Okay, wait, hold on.
- 3:23SPEAKER_01: I actually want to push back on that margin clarity for a second.
- 3:26SPEAKER_00: Oh, okay.
- 3:27SPEAKER_00: What's up?
- 3:27SPEAKER_01: Because there's this specific counterintuitive effect reported in the latest quarter
- 3:32SPEAKER_01: about high bandwidth memory, HBM, that literally makes zero sense to me.
- 3:37SPEAKER_00: Ah, I know exactly what you're talking about.
- 3:39SPEAKER_00: The dilution effect.
- 3:41SPEAKER_01: Yes.
- 3:42SPEAKER_01: HBM is their absolute premium product, right?
- 3:45SPEAKER_01: Everyone is fighting for it.
- 3:46SPEAKER_01: Yet the filing says
- 3:47SPEAKER_01: that selling HBM actually diluted Cloud Memory's blended gross margin this quarter.
- 3:53SPEAKER_01: How does selling more of your best product drag your profit margin down?
- 3:57SPEAKER_00: I know, it sounds totally backwards.
- 4:00SPEAKER_00: But this is purely a product-sales mix effect.
- 4:03SPEAKER_01: Okay, explain that to me literally.
- 4:05SPEAKER_00: So memory products differ,
- 4:07SPEAKER_00: and HPM's gross margin was mathematically lower
- 4:10SPEAKER_00: than conventional DRAM's gross margin this specific quarter.
- 4:14SPEAKER_01: Oh, I see.
- 4:16SPEAKER_00: Right.
- 4:16SPEAKER_00: So if you hold everything else equal,
- 4:19SPEAKER_00: a higher share of lower margin HPM sales is going to naturally lower
- 4:23SPEAKER_00: the blended gross margin for cloud memory overall.
- 4:26SPEAKER_01: Just because the product itself costs more to make right now relative to its price.
- 4:30SPEAKER_00: Exactly.
- 4:31SPEAKER_00: Even as overall pricing improved, that mix pulls the average down.
- 4:36SPEAKER_00: Now, they do expect higher agreed HBM prices for 2027 to narrow this gap.
- 4:41SPEAKER_01: Okay, that makes sense as a math equation.
- 4:44SPEAKER_01: But what about the factory side of things?
- 4:46SPEAKER_00: Right.
- 4:46SPEAKER_00: So that's the other crucial piece.
- 4:48SPEAKER_00: The use of actual wafer capacity on the factory floor is
- 4:52SPEAKER_00: a completely different industry supply mechanism.
- 4:54SPEAKER_01: Meaning it's physical, not just an accounting mix.
- 4:57SPEAKER_00: Right.
- 4:58SPEAKER_00: Choosing to print HBM instead of conventional DRAM is a physical supply choice.
- 5:02SPEAKER_00: It is not some booked cost of foregone volume that explains that reported margin mix effect.
- 5:08SPEAKER_00: They're two different things.
- 5:10SPEAKER_00: Got it.
- 5:11SPEAKER_01: Okay, so moving from earnings to the balance sheet,
- 5:14SPEAKER_01: because,
- 5:14SPEAKER_01: I mean, the earnings are sky high, but where is all that cash actually sitting?
- 5:18SPEAKER_00: Well, at first glance, the liquidity is incredible.
- 5:21SPEAKER_00: They have about $73 billion of cash in investments.
- 5:24SPEAKER_01: Yeah, against only, what, about $5.2 billion of debt?
- 5:27SPEAKER_00: Right, including their finance leases.
- 5:29SPEAKER_00: But there's a huge caveat here.
- 5:32SPEAKER_01: There's always a caveat.
- 5:33SPEAKER_00: Always.
- 5:34SPEAKER_00: About $13 billion of that cash balance consists of received customer security deposits.
- 5:39SPEAKER_01: Wait, $13 billion is just deposits?
- 5:42SPEAKER_01: Like a security deposit on an apartment?
- 5:44SPEAKER_00: Basically, yeah.
- 5:46SPEAKER_00: Mechanically, these are a finance and cash flow classification.
- 5:49SPEAKER_00: They are refundable contract liabilities.
- 5:52SPEAKER_01: Okay, so it's not cash they earned just by delivering chips.
- 5:55SPEAKER_00: Exactly.
- 5:56SPEAKER_00: It's not revenue yet.
- 5:58SPEAKER_00: Micron has to return that deposit when
- 6:00SPEAKER_00: the customers actually meet their minimum purchases down the line.
- 6:04SPEAKER_01: Which is scheduled for when?
- 6:06SPEAKER_00: Nearly all of those repayments are scheduled for 2029 through 2031.
- 6:09SPEAKER_00: Wow, okay.
- 6:12SPEAKER_01: And what if the customer bails?
- 6:14SPEAKER_01: What if they don't buy the chips?
- 6:15SPEAKER_00: If the customer fails to meet the commitments, then Micron may retain the deposit.
- 6:19SPEAKER_00: But until then, that whole $73 billion cash balance is simply not free to distribute.
- 6:25SPEAKER_01: Right.
- 6:25SPEAKER_01: You can't pay a dividend with someone else's security deposit.
- 6:28SPEAKER_00: Exactly.
- 6:29SPEAKER_01: And they need that cash, right?
- 6:30SPEAKER_01: Because they are spending an unbelievable amount of money on these new factories.
- 6:35SPEAKER_00: The investment cycle is massive.
- 6:36SPEAKER_01: Yeah.
- 6:36SPEAKER_01: Looking at the fiscal 2027 first half,
- 6:40SPEAKER_01: they have $25 billion in total net capital expenditure just in the first half.
- 6:45SPEAKER_00: Right.
- 6:45SPEAKER_00: And we need to clarify that CapEx figure because a common misconception is
- 6:49SPEAKER_00: that this $25 billion is all construction.
- 6:52SPEAKER_01: It's not.
- 6:53SPEAKER_00: No, it's not all construction.
- 6:54SPEAKER_00: While a majority of the increase when you compare
- 6:57SPEAKER_00: to their prior fiscal 2027 plans is construction,
- 7:01SPEAKER_00: specifically accelerating clean room space for late 2028 and beyond.
- 7:05SPEAKER_01: Okay.
- 7:05SPEAKER_01: So a lot of the new spending is the physical buildings.
- 7:08SPEAKER_00: Right.
- 7:09SPEAKER_00: But equipment investment remains in that total figure.
- 7:11SPEAKER_00: And some of it was pulled forward.
- 7:13SPEAKER_01: Gotcha.
- 7:14SPEAKER_01: Gotcha.
- 7:14SPEAKER_00: And mechanically,
- 7:15SPEAKER_00: spending cash now to build a factory and then recognizing the depreciation later on
- 7:20SPEAKER_00: the income statement are very different financial effects.
- 7:23SPEAKER_01: So how do we bridge all this?
- 7:25SPEAKER_01: The operating cash, the CapEx, the deposits?
- 7:28SPEAKER_01: That's a lot of moving parts.
- 7:29SPEAKER_00: We can tie it together with one cash bridge.
- 7:31SPEAKER_00: You start with about $90 billion of operating cash.
- 7:35SPEAKER_01: Okay.
- 7:35SPEAKER_00: Then you minus about $31 billion of gross capital spending.
- 7:39SPEAKER_00: That leaves you with $59 billion.
- 7:41SPEAKER_00: Right.
- 7:41SPEAKER_00: $59 billion.
- 7:43SPEAKER_00: Then with asset sales and capital incentives added back in,
- 7:47SPEAKER_00: the company defined adjusted free cash flows about $62 billion.
- 7:51SPEAKER_00: Okay.
- 7:52SPEAKER_01: And just to be clear, those customer deposits we talked about, those are financing cash, right?
- 7:57SPEAKER_00: Yes.
- 7:58SPEAKER_00: They are already outside both the operating cash and this $62 billion adjusted measure.
- 8:03SPEAKER_01: Right.
- 8:04SPEAKER_01: You know,
- 8:05SPEAKER_01: looking at this massive capital outlay,
- 8:07SPEAKER_01: it brings to mind our actual October 23 Amazon conversation.
- 8:10SPEAKER_00: Oh, that's a great parallel.
- 8:11SPEAKER_01: Yeah.
- 8:12SPEAKER_01: And just to be super clear for you listening,
- 8:14SPEAKER_01: this comparison is solely about the capital allocation question faced
- 8:17SPEAKER_01: by two different businesses.
- 8:19SPEAKER_00: Right.
- 8:20SPEAKER_00: Whether heavy upfront investment actually earns durable cash returns over time.
- 8:24SPEAKER_01: Exactly.
- 8:25SPEAKER_01: It is not an assertion about direct purchases or customer concentration
- 8:28SPEAKER_01: between Amazon and Micron.
- 8:30SPEAKER_01: They just both have to answer that same tough question.
- 8:33SPEAKER_00: It's the core puzzle for both of them.
- 8:35SPEAKER_01: And by the way,
- 8:36SPEAKER_01: if you find this capital investment puzzle interesting,
- 8:38SPEAKER_01: you should definitely check out our earlier NVIDIA and Apple deep dives too.
- 8:42SPEAKER_01: We explore some really different perspectives on valuation there.
- 8:45SPEAKER_00: Those are fantastic companions to this discussion.
- 8:48SPEAKER_01: Yeah, they really are.
- 8:50SPEAKER_01: So getting back to Micron, I want to introduce a separate return illustration.
- 8:54SPEAKER_01: We talked about the $878 market close that the current ratings use.
- 8:58SPEAKER_00: Right.
- 8:58SPEAKER_00: The official consensus basis.
- 9:01SPEAKER_01: But I want to look at a dated after hours trade from October 23,
- 9:05SPEAKER_01: specifically at 4.56 p.m. New York EDT, which was at $880.19.
- 9:10SPEAKER_01: Okay, $880.19.
- 9:15SPEAKER_01: Yes.
- 9:16SPEAKER_01: Can you explain the base path from this specific reference?
- 9:19SPEAKER_01: And keep in mind, this is explicitly different from that $878 market close.
- 9:24SPEAKER_00: Sure.
- 9:24SPEAKER_00: So based on the deep research, we have fixed endpoints.
- 9:27SPEAKER_00: The targets are about $996 for the third quarter of 2027 and $1001 for
- 9:35SPEAKER_00: the third quarter of 2031.
- 9:36SPEAKER_00: Okay.
- 9:38SPEAKER_00: So from that 880.19 reference,
- 9:40SPEAKER_00: those targets imply cumulative price gains of about 13.1% and 13.7%.
- 9:47SPEAKER_01: But we also have to factor in dividends, right?
- 9:49SPEAKER_01: Micron pays a dividend of 15 cents quarterly, which is 60 cents annually.
- 9:53SPEAKER_00: We do, but we have to be precise with the dates.
- 9:56SPEAKER_01: Right.
- 9:56SPEAKER_00: So we exclude the October 29 payment because its October 14 record date has already passed.
- 10:01SPEAKER_01: Makes sense.
- 10:03SPEAKER_00: So using a flat payout,
- 10:04SPEAKER_00: we model three payments, which is 45 cents, before that first 2027 endpoint.
- 10:10SPEAKER_01: Okay.
- 10:10SPEAKER_01: And for the 2031 endpoint?
- 10:13SPEAKER_00: We model 19 payments, which is about $2.9 before the second endpoint.
- 10:18SPEAKER_01: So when you add that cash back in,
- 10:20SPEAKER_01: that gives us roughly 13.2%
- 10:24SPEAKER_01: and 14.0% cumulative gross cash-inclusive returns.
- 10:28SPEAKER_00: Exactly.
- 10:29SPEAKER_00: But, you know, those flat endpoints completely conceal a very uneven path to get there.
- 10:34SPEAKER_01: Oh, for sure.
- 10:35SPEAKER_01: It's never a straight line.
- 10:37SPEAKER_01: What are the forces that could actually disrupt this trajectory?
- 10:39SPEAKER_00: Well, you have ordinary forces first.
- 10:42SPEAKER_00: The drivers are things like content growth, waiver constraints, and agreements.
- 10:47SPEAKER_00: And the frictions?
- 10:48SPEAKER_00: The ordinary frictions are that massive construction bill we discussed.
- 10:51SPEAKER_00: The fixed depreciation and competing supply.
- 10:54SPEAKER_00: But then, there are two distinct tails.
- 10:58SPEAKER_01: The extremes.
- 10:59SPEAKER_00: Right.
- 10:59SPEAKER_00: The upside tail is that rival packaging remains difficult for longer than expected.
- 11:04SPEAKER_00: And new AI uses just absorb all the capacity.
- 11:07SPEAKER_01: Meaning Micron extends its pricing power.
- 11:10SPEAKER_00: Exactly.
- 11:11SPEAKER_00: But the downside tail is a sharp AI spending retrenchment plus new supply.
- 11:17SPEAKER_01: Which would be a disaster.
- 11:19SPEAKER_00: Yeah.
- 11:19SPEAKER_00: Because in plain analytical language,
- 11:21SPEAKER_00: if HBM demand weakens,
- 11:24SPEAKER_00: suppliers could redirect their wafer capacity back to conventional DRAM.
- 11:28SPEAKER_01: Oh, so they just switch the machines over.
- 11:30SPEAKER_00: You know, it's a conditional industry mechanism, not an instant conversion.
- 11:33SPEAKER_00: But yes, it would intensify a price shock across the whole market.
- 11:37SPEAKER_01: That perfectly leads into the alpha gap hypothesis from the research.
- 11:41SPEAKER_00: The idea that exceptional peak earnings can conceal weaker through cycle economics?
- 11:46SPEAKER_01: Exactly.
- 11:46SPEAKER_00: Right.
- 11:47SPEAKER_01: It's basically asking if this boom is blinding us to the long-term reality.
- 11:51SPEAKER_01: And it's causing a major debate among the advisors on the iPulse AI platform.
- 11:55SPEAKER_00: It really is.
- 11:56SPEAKER_00: You have optimistic perspectives saying that custom co-design
- 11:59SPEAKER_00: and take-or-pay agreements make this memory cycle durable.
- 12:02SPEAKER_01: Right.
- 12:03SPEAKER_01: That it's different this time.
- 12:04SPEAKER_00: Right.
- 12:05SPEAKER_00: But then you have skeptical perspectives arguing
- 12:07SPEAKER_00: that these high margins are just funding the next wave of supply.
- 12:10SPEAKER_01: Which brings prices crashing down eventually.
- 12:13SPEAKER_00: And I should mention for you listening, these are AI generated perspectives on the platform.
- 12:18SPEAKER_00: They may share underlying models and data.
- 12:20SPEAKER_00: They are not independent human analysts.
- 12:22SPEAKER_00: That's a good point.
- 12:24SPEAKER_00: And it's interesting to see how the overall narrative shifted over the summer.
- 12:27SPEAKER_01: Yeah, the change from July to September was huge.
- 12:30SPEAKER_00: It was a real business change.
- 12:32SPEAKER_00: In July, the focus was mostly on a generic late 2027 memory glut.
- 12:36SPEAKER_01: But by September...
- 12:37SPEAKER_00: By September, it became very specific.
- 12:39SPEAKER_00: It was suddenly a contest between contractual resilience versus delayed capacity expansion.
- 12:45SPEAKER_01: Meaning the major supply wave risk was pushed out toward 2028.
- 12:49SPEAKER_00: Exactly.
- 12:51SPEAKER_00: And the models project this timeline with weighted paths.
- 12:54SPEAKER_01: Let's hear the actual forecast numbers.
- 12:56SPEAKER_00: They model a peak roughly at $1,109 in the first quarter of 2027.
- 12:59SPEAKER_00: That's a 26% gain.
- 13:00SPEAKER_00: Okay, a nice peak.
- 13:00SPEAKER_00: And the trough.
- 13:00SPEAKER_00: The trough is around $657 in the fourth quarter of 2028,
- 13:02SPEAKER_00: which is down 25%
- 13:03SPEAKER_01: before recovering to the 2031 endpoint.
- 13:16SPEAKER_01: But we need to be really careful here because those turning points in
- 13:19SPEAKER_01: the forecast don't actually line up with the corporate dates for the new factories,
- 13:23SPEAKER_01: do they?
- 13:23SPEAKER_00: No, they don't.
- 13:24SPEAKER_00: And we have to caution against assuming causal alignment there.
- 13:28SPEAKER_01: Right.
- 13:29SPEAKER_00: The actual dates are the first Idaho output is mid-2027,
- 13:34SPEAKER_00: the second is late-2028, and New York is in 2030.
- 13:39SPEAKER_00: The price path turning points happen independently of those physical factory dates.
- 13:44SPEAKER_01: Because the market anticipates the supply before the factory actually opens its doors.
- 13:49SPEAKER_00: Precisely.
- 13:50SPEAKER_01: Okay, so this has been a massive amount of data.
- 13:53SPEAKER_01: What is the ultimate investment question here?
- 13:56SPEAKER_01: Like what evidence would actually prove this view wrong?
- 13:59SPEAKER_00: Well, we look for plain, literal tests.
- 14:02SPEAKER_00: If we see persistently strong cash generation through growing supply,
- 14:06SPEAKER_00: that weakens the skeptical thesis.
- 14:08SPEAKER_01: It shows the earnings are durable.
- 14:09SPEAKER_01: Right.
- 14:10SPEAKER_00: Right.
- 14:10SPEAKER_00: But on the flip side,
- 14:11SPEAKER_00: if we see slower pricing,
- 14:13SPEAKER_00: weaker orders, or rising inventory, that directly weakens the optimistic case.
- 14:18SPEAKER_01: Okay, so to synthesize all this for you listening,
- 14:21SPEAKER_01: you really have to treat the bullish mechanisms and the cyclical risks together.
- 14:25SPEAKER_00: You can't just pick one.
- 14:27SPEAKER_01: Right.
- 14:27SPEAKER_01: Don't just assume permanent scarcity, but also don't assume an inevitable crash.
- 14:32SPEAKER_00: It's a balance.
- 14:33SPEAKER_01: Definitely.
- 14:34SPEAKER_01: And we invite you to explore the Micron consensus research,
- 14:38SPEAKER_01: the AI opinions,
- 14:39SPEAKER_01: and the profile and fundamentals on iPulse AI to inspect this evidence for yourself.
- 14:44SPEAKER_00: Highly recommend looking at the data directly.
- 14:46SPEAKER_01: And as we wrap up, I just want to leave you with a thought.
- 14:49SPEAKER_01: When you see a company pulling in massive profits, ask yourself, are they building a moat?
- 14:54SPEAKER_01: Or are they just funding the factories that will eventually flood their own market?
- 14:58SPEAKER_01: Like, comment, and subscribe.
- 14:59SPEAKER_01: We are publishing a new podcast every day of the week except Thursday.
- 15:03SPEAKER_01: Stay tuned for the next podcast on Adyen.
- 15:05SPEAKER_01: How does a payments platform turn merchant growth into durable earnings?