BING-JHEN HONG
In this memory upcycle, I’ve been more accurate in locating the fundamental turns than at any other time in my analysis career. After pointing out the exact industry and share bottom in December of 2022 when Micron Technology, Inc. (NASDAQ:MU) said bits shipped had bottomed, many of my subscribers (and maybe even some of my followers) entered into large common share or long-dated call (LEAP) long positions to capitalize on the forthcoming upcycle. Then, several months ago, when I heard management’s plan for CapEx and fab equipment, I predicted this would be the best upcycle in history in terms of producers’ benefit. Now, after a stellar FQ2 earnings report and a 150% return on my 2022 call, it’s time to reassess once again where we are in the fundamental and financial cycle and determine the risk-reward of the share price.
A Brief History Of How We Arrived Here
The last year-and-a-half has been an interesting ride for the Micron crowd. In the weeks and months following my December 2022 article, many disagreed with me, saying there was much more to go in the cyclical trough, while others said earnings were still bottoming and there’d be better entries to come in 2023. But those who don’t understand how Micron’s industry and cycles work weren’t going to be convinced the financials don’t necessarily matter when finding the bottom (or the top, for that matter).
The most significant factor in investing is finding the best share entry, and to do so for Micron means watching the industry’s vitals and, more specifically, shipments. This is why bits shipped bottoming for Micron was the primary data point needed to determine the bottom in the industry and the stock.
Once bits shipped began their uptick, DRAM and NAND pricing followed. Because pricing followed, bits shipped have now taken a pause as seasonality and orders line up with higher prices in mind. But, more directly, it’s because supply has and continues to decrease due to actions taken in mid-to-late 2022 and all of 2023. It’s difficult to ship more bits when you’re intentionally producing less, so the relative change of bits shipped will not grow sequentially for the entirety of the upcycle.
But moving toward more recent events on the back of my “this time is different” upcycle article, some have turned to market share and technology, wringing hands over Micron’s HBM3’s (high bandwidth memory generation 3 “slow start” among other things. This, again, proved fruitless, and the same thinking that led to missing the stock’s rally over the last year led to calling for taking profits ahead of last month’s earnings report, missing some 30% upside post earnings.
Picking The Right Points To Focus On
The problem with many attempting to value Micron, its business, and its stock is they’re looking at the trees, not the forest. Micron’s individual business decisions don’t help understand where the share price is going. This is why many got in late and bailed early, and some still think the upcoming year will be disappointing and recent results are an illusion.
Supply: It’s Just Not There, Neither Is Price
By waiting until HBM3e ramps up its HBM program, Micron has a smaller HBM market share. But it means its market share can only go up from here. Meanwhile, SK Hynix, the HBM market share leader today, must now defend its market share in the face of an upcycle. Micron will not only gain market share but also add to the top and bottom line through HBM’s overall market expansion. SK Hynix will lose market share during the next two years, but it’ll grow HBM revenues through overall market growth.
I’d rather be in Micron’s position.
But this isn’t the complete picture.
Because the DRAM players are reallocating equipment toward HBM, they’re doing so through prior underutilized equipment, something I laid out extensively in my “best upcycle” article. This means the supply of other DRAM disciplines and end markets is structurally eliminated. And because HBM requires three times as many wafers to produce the equivalent amount of bits, the reallocation takes material wafer supply offline, different than shifting it. Therefore, supply is hurtling toward levels from years ago, undoing any capacity expansion over the last several years.
Industrywide, HBM3E consumes approximately three times the wafer supply as D5 to produce a given number of bits in the same technology node. With increased performance and packaging complexity, across the industry, we expect the trade ratio for HBM4 to be even higher than the trade ratio for HBM3E.
– Sanjay Mehrotra, CEO, FQ2 ’24 Earnings Call.
This means supply will be severely limited this coming year, and DRAM and NAND prices will rise considerably over the next several quarters on top of what it already has. In other words, unless demand suddenly drops to 2021 levels, the supply-demand equation won’t be balanced anytime soon.
You need to understand there’s never been a time where we’ve seen a decrease in wafer output outside of temporary margin-hurting underutilization. Node transitions in the past have spurred some net-zero gains in wafer output, but between at-then strategies of adding more lines to make up for lost wafer output and bit size shrink to turn out more bits per wafer, bit supply ultimately slowly increased, even if net wafers didn’t. But taking lines offline – and thus wafers – and allocating them to end products requiring three times those wafers means massive, permanent reductions in wafer and bit output.
This leads to a supply constraint environment and, thus, price increases across the board. After seeing DRAM pricing increase at a high-teens percentage rate quarter-over-quarter, the company expects FQ3 to see continued strength in pricing.
…we expect solid sequential improvement in fiscal Q3 gross margins due to robust price increases across both DRAM and NAND.
– Mark Murphy, CFO, FQ2 ’24 Earnings Call.
The upcycle is just getting started, and earnings will continue to grow quarter-over-quarter for several quarters, leading to “record revenue in fiscal 2025,” according to the CFO. The share price is coming up to speed with this as bullish sentiment carries it in this first leg of the rally.
Supply down, pricing up.
That’s what you want to see to turn a sour memory market into a thriving one for producers. And supply just isn’t there to outdo demand. But pricing isn’t yet, either; there’s still more to run.
Inventory Is Still High
An upcycle at its peak can be characterized by lean inventory levels. When demand outstrips supply, finished goods are being shipped from every door, and raw materials are being used faster than they can be ordered. At this point, inventory is still at very elevated levels.
Management is targeting 120 days of inventory outstanding. Currently, DIO stands at 160 days, the same as FQ1. Without doing any math, you can see it didn’t decrease. Lest you think DIO is mostly accounting engineering, in absolute dollar terms, it was up slightly year-over-year and quarter-over-quarter.
Micron’s FQ2 ’24 Press Release
Therefore, inventory has a ways to go. Management expects it to track down for the next year and a half.
We expect to reduce inventory levels and excluding strategic inventory stock, be within a few weeks of our 120 days target by the end of fiscal 2024. We project DIO improvements to continue into fiscal year 2025.
– Mark Murphy, CFO, FQ2 ’24 Earnings Call.
This precludes the top from being in, as inventory leans out before the top is put in and maxes as the bottom hits. But it’s not a true tell-tale sign; it’s just another characteristic, as the “top” or “bottom” of inventory isn’t always aligned with the true top or bottom. Inventory today is giving us indications there are still several quarters before it flashes warning signs.
However, because there’s a fundamentally different approach to CapEx and supply during this upcycle, there may not be an exact alignment of inventory and the cycle stages. In the past, when times were good, more CapEx investments were going toward increasing supply. This was met with a shift in the supply and demand balance, either through increasing supply, a steep fall in demand, or both. That added capacity would then get shoved into inventory, and the cycle shift would become apparent. Since supply investments are being pulled back significantly to the point where bit output was down year-over-year – yes, negative in 2023 – the only avenue for inventory to increase is demand slows to below the now trending down supply.
The conclusion is inventory may be something to keep an eye on, but I wouldn’t count on it this time around, especially considering it was never a great sign of the top or bottom anyway. Going forward, it may be even less of a data point, as it won’t provide predictive power and may stay lean for longer than historical trends suggest.
It’s Technological Position
Micron has been proving doubters wrong for years regarding memory manufacturing and the technology therein. This is credited to Sanjay Mehrotra’s leadership, which started in 2017. Since then, the company has leapt ahead of its major competitors in DRAM and NAND node technology. However, one thing has always stood out for the company: it never followed competitors’ roadmaps or strategies.
This isn’t truer than its decision to forgo EUV lithography for several nodes while its peers switched to it nearly immediately after hitting the Greek nodes. Due to this, many wondered if Micron would continue its leadership in engineering and cost-competitiveness.
As it turns out, DUV made Micron’s life much easier than its Korean counterparts, as it, one, didn’t have to purchase expensive equipment upfront, and, two, didn’t have to deal with the infancy of EUV for memory fabrication and the growing pains of different lithographic shifts to be compatible with the new paradigm.
However, what was slightly unexpected was the details on mirroring its alpha and beta node lines on the EUV equivalent flow and how it matched quality and yield. This isn’t something to gloss over. Many believed Micron would be behind in the EUV space because it would have the least experience, as its competitors had it in production for several years before Micron’s intersection in production. Instead, Micron appears to not have missed a beat.
We continue to mature our production capability with extreme ultraviolet lithography and have achieved equivalent yield and quality on our 1-alpha as well as 1-beta nodes between EUV and non-EUV flows. We have begun 1-gamma DRAM pilot production using EUV and are on track for volume production in calendar 2025.
– Sanjay Mehrotra, CEO, FQ2 ’24 Earnings Call (emphasis added).
Micron shouldn’t have any issue intersecting its EUV program with its 1-gamma node and the nodes to follow with this kind of early success.
You may ask how this helps in an upcycle. Well, when supply is tight, every bit off the line is another sale, and good yields are king. While its competitors are beyond their transition, if Micron can complete its crossover smoothly and with fewer yield issues, it can catch up in market share at a time when yield and cost leadership are king.
It’s like in a race where one team has a strategy to pit on lap 30, while your team’s strategy is to pit on lap 35. While you didn’t come in at the same time, the difference in time on pit road matters. If you were two seconds faster pitting, it shows up in track position in the race following the pit stop. It could mean overtaking them because you came out of the pits ahead of them as they were still coming out of turn four. You didn’t have to be on pit road at the same time for it to make a difference on the track.
Some have expressed concern about this ever since the company decided to continue with DUV beyond its competitors. With this confirmation of equivalent yields, Micron’s ability to maintain leadership and produce mature yields will provide upside in the upcycle during a large lithographic technological shift.
Now, this technological leadership won’t help determine when the top of the cycle is in, but it makes this upcycle stronger. Fewer execution hiccups will lead to very efficient and record-breaking earnings, on top of potential gains in market share.
“Real” Contracts This Time
A “this time is different” factor contributing to a stronger upcycle is due to HBM’s massive demand for AI accelerators and related products. With such a small supply across the industry, securing it puts producers in the catbird seat. And this isn’t the “supply contracts” of years past that seemed not to mean anything when inventory digestion and end market demand declined; they all went out the window. But it appears now Micron has firm volume and price contracting, at least for HBM.
…when we talk about that HBM is sold out, those type of contracts have both pricing as well as volumes as well as other stricter terms baked in as part of our LTAs. And 2024 volume as well as pricing is all locked up.
2025, as I mentioned, the volumes are largely allocated. A vast majority of our production supply is allocated, and some of the pricing is already firmed up.
– Sanjay Mehrotra, CEO, FQ2 ’24 Earnings Call Q&A (emphasis added).
This has led to being “sold out” of HBM3e for 2024, with much of the 2025 supply already allocated. This is a new phenomenon in the memory industry, at least for Micron. Never before have volume and price been placed in a long-term contract four quarters out. This speaks to the insatiable demand for HBM and AI accelerators.
Moreover, this demand-pull on HBM is causing supply constraints in non-HBM DRAM, as supply continues to be diverted to the higher-margin HBM product. Perhaps if things get bad enough in non-HBM land, these types of contracts could spill over into the less niche memory types. For now, having firm contracts with volume and price negotiated allows Micron to have great visibility for orders but also CapEx and supply planning.
So, What’s The New Outlook?
Is it different this time? Well, you tell me. For one, the MU stock price is deep into all-time high territory, running up to $130 this week. Two, the stock is at new highs well before the peak of earnings. I figure we’re still four or five quarters away from a slowdown in earnings, with the peak likely not too far beyond that. But, as I’ve detailed in articles from years ago, the stock will fall before the earnings reports show declining sequential growth. 2022 was a great example, as the stock was down some 43% from its peak by the time peak earnings were reported. So, with that in mind, we may still have a few more quarters of returns to go.
But from a valuation and chart perspective, the Micron Technology, Inc. reward is closing, and the risk is widening. With my pounding-the-table-buy call when it was under $50, trimming at $125-$130 is prudent, especially for Micron. I followed my own advice and completed my first ever trim, MU shares I’ve had since 2015. It was a single-digit percentage of the holding, but a trim nonetheless.
From a valuation perspective, it makes sense as well.
Book value is the most accurate way to value Micron. Price-to-earnings is hardly meaningful with the amount of changes and swings it sees in earnings. Even then, the general rule of thumb is buy when the P/E is high and sell when it’s low. However, price-to-book value is a better, though not perfect, method. Typically, when it’s near 1, you buy (see my $49 buy call when it was near 1), and when it nears 3, you sell.
The only thing is, due to the rapid rise in share price ahead of the cycle getting going (so while book value was decreasing), the price-to-book value may be skewed for a higher ceiling than history would dictate. This creates a bit of a conundrum in terms of whether it’s the peak of the share price while the valuation mellows out or the peak of both. In other words, the slight relative decrease in book value caused the price-to-book ratio to rise simultaneously with the stock rise, but the increase in book value expected going forward should compress the valuation. The question is, did the share price peak or just the valuation ratio?
From a chart perspective, I see more upside ahead, but the window is closing from a risk/reward aspect. I expect a consolidation in the low $100s, followed by one more push in this upcycle to $150 or so. From there, I’d consider the cycle’s rally complete from a technical chart perspective. With this in mind, combined with the potential 20% reward to my $150 target from current prices, I’m moving from a buy to a hold on the stock. Consider the move to $100 as one more buying opportunity before a 50% run.
The conditions in the memory industry are as good as they can be – in fact, better than ever – at this stage. But the stock has run up quite quickly ahead of record revenue – and likely record earnings – and with the valuation at historical upper-level ranges, buying here would be riskier than even just a few months ago, let alone my late 2022 article. I wasn’t in favor of moving to hold in the $90s, but with the nice 30% run post-earnings, it’s time to let things settle out. After all, the share price matters in the end; it’s how your investments make (or lose) money. It’s not necessarily the quibbles over Micron’s market share and differing technology strategies and roadmaps. But some will still say it is.
