Micron Is Up 230% on the Memory Supercycle—Here's Who's Losing
Samsung just posted 89 trillion won ($64 billion) in quarterly semiconductor operating profit — and lost money on smartphones for the first time in its history. Same company, same quarter, two completely opposite outcomes. That split isn't a Samsung problem. It's the clearest read yet on where the memory supercycle is creating winners and where it's quietly wrecking margins.
Photo by Sergei Starostin on Pexels
The Numbers Behind the "Supercycle" Word
This isn't hype-speak. DRAM contract prices are on pace to rise roughly 125% for full-year 2026, and NAND flash prices are tracking an even steeper 234% increase, according to industry pricing data compiled by Utmel's 2026 memory market analysis. Q1 alone saw conventional DRAM jump 90-95% quarter-over-quarter. That kind of move doesn't happen in a normal cycle — it happens when hyperscalers building AI data centers are buying up capacity that used to go to laptops and phones.
Micron (MU) is the purest US-listed way to play that. As of late July, the stock was trading around $921, up roughly 230% year-to-date, with its HBM4 high-bandwidth memory now in volume production for AI accelerators. That's the winning side of the trade, and the market has already noticed — loudly.
The Losing Side Nobody's Pricing In Yet
Here's what's less obvious: the same shortage that's minting profits for memory makers is actively squeezing the companies that buy memory to build finished products. Samsung's mobile division (MX) swung to a roughly 700 billion won loss this quarter — its first ever — specifically because component costs, memory chief among them, outran what it could pass on to customers.
Apple isn't immune either. Reports from MacRumors in February indicate Apple agreed to a 100% price hike on Samsung memory chips for its devices, and Tim Cook has already flagged that rising chip costs will have "a bit more of an impact" on gross margins going forward. When the world's highest-margin phone maker is warning about this on an earnings call, it's not a rounding error.
The damage is worse further down the food chain. According to IDC's 2026 outlook, global smartphone shipments are projected to fall 12.9% this year — the industry's steepest decline on record — largely because memory and storage now eat up more than 60% of total production cost on budget phones, versus roughly 30% on premium devices. Budget and mid-tier phone makers have almost nowhere to hide; premium players like Apple and Samsung can absorb the hit for now because their overall margins are fatter, but "can absorb it" and "won't feel it" are not the same thing.
Why This Cycle Doesn't Look Like the Last One
Memory has always been cyclical — brutal downturns follow every boom, which is exactly why Samsung and SK Hynix stock often gets punished even during profit spikes, something I dug into in SK Hynix Posts Record Profit, Stock Sinks 8.76% on the Miss. What's different this time is the demand source. Historically, memory demand followed PC and phone replacement cycles — predictable, seasonal, and prone to overbuilding. This cycle is being driven by Meta, Google, and Microsoft racing to secure HBM and DRAM capacity for AI infrastructure, and manufacturers are openly prioritizing that higher-margin business over consumer electronics.
That's structurally different because AI capex isn't slowing down on a normal replacement schedule — it's tied to a buildout that companies have said will run for years. Analysts quoted in the same IDC report don't expect memory prices to meaningfully ease before the second half of 2027, and some think pricing never fully reverts to pre-2025 levels. If that holds, this isn't a spike investors should expect to fade quickly — which is also why I flagged the shortage narrative as a multi-quarter story rather than a one-off in Samsung's 2027 Shortage Warning Sparks Kospi's Sharpest Rebound.
So Is Micron Still Investable at +230%?
The honest answer: the easy money on "memory prices are going up" has been made. A stock up 230% year-to-date has already priced in a lot of good news, and any hint that AI capex growth is decelerating — even a deceleration in the rate of growth, not an outright decline — could hit MU harder than a company trading at a more modest multiple. That's the real risk here, not whether the supercycle is real.
What would make me more cautious on Micron specifically:
- Any signs hyperscaler capex guidance is being trimmed for 2027
- Inventory buildup signals from PC/phone OEMs suggesting they're pulling back orders at these price levels
- A sudden capacity add announcement from Samsung or SK Hynix that could ease the shortage faster than expected
On the other side of the trade, the margin-squeeze story on device makers is arguably less discovered than the Micron rally. Apple's stock hasn't meaningfully reflected the "chip costs are eating into margins" narrative yet — Cook's own comments were fairly muted, and the market has been more focused on iPhone unit demand than component cost inflation. If NAND keeps climbing at anywhere close to the pace TrendForce and IDC are projecting, that's a headwind that shows up in gross margin lines over the next few quarters, not something priced in today.
The Takeaway
The memory supercycle is real, verifiable, and not close to over — but the trade has bifurcated. Chasing MU here means betting AI capex keeps accelerating faster than the stock's run-up already assumes. The more interesting, less crowded angle is watching how device makers like Apple manage margin compression over the next two to three earnings cycles, since that's the side of this story the market hasn't fully priced yet. This is not financial advice — always do your own research before making investment decisions.
댓글
댓글 쓰기