Seagate Leads Memory Sell-Off as CEO Says It Would ‘Take Too Long’ to Build New Factories, What Investors Need to Know
Seagate Leads Memory Sell-Off as CEO Says It Would ‘Take Too Long’ to Build New Factories, What Investors Need to Know
The Quiet Admission That Shook a $100 Billion Sector
Sometimes the most market-moving statements aren’t shouted, they slip out in a conference Q&A, almost as an aside.
That’s exactly what happened on Monday, May 18, 2026, when Seagate Technology CEO Dave Mosley took the stage at the JPMorgan Global Technology, Media and Communications Conference. Someone asked him a fairly routine question: Is Seagate planning to build new factories to keep up with this massive AI-driven demand?
Mosley’s answer was refreshingly candid, and it triggered a sell-off that erased roughly $100 billion in market value across the memory chip sector within hours.
“If we took the teams off and started building new factories or bringing up new machines,” he said, “that would just take too long.” (And then, almost as if thinking out loud:) “You would end up with more capacity, but then you’d slow the rate of growth on that technology.”
Seagate shares tumbled as much as 8%, eventually closing down about 7.4% at $736.89, a stark drop from the prior close of $795.47. But the pain didn’t stay contained. Micron Technology fell nearly 6%. Western Digital and SanDisk each slid around 7%. Nvidia, Broadcom, and Intel, AI darlings with tangential exposure, all dipped in sympathy.
The market heard one word: supply can’t keep up.
But here’s the thing, Mosley wasn’t admitting failure. He was explaining a deliberate strategy. And if you’re willing to look past the panic, there’s a much more interesting story here.
What Actually Happened at That JPMorgan Conference
Let’s reconstruct the moment, because context matters.
Mosley was speaking at 9:45 AM ET on May 18 at one of the most closely watched investor conferences on the calendar. The room was packed with analysts, portfolio managers, and sector specialists, all of whom have watched memory chip stocks soar anywhere from 120% to 380% so far in 2026.
The question on everyone’s mind: demand for AI storage is clearly surging. So why not just build more factories?
Mosley explained that Seagate’s critical components, particularly recording head wafers, already have lead times exceeding nine months. After those wafers arrive, it takes another full quarter to manufacture the finished drives. That’s a year-long pipeline from component order to finished product, before you even think about constructing new factory floor space.
He then outlined the company’s actual growth strategy: targeting mid-20s percent compound annual growth (CAGR) by advancing through technology transitions, specifically, moving from 3-terabyte-per-platter drives to 4TB and eventually 5TB per platter. The company’s Mozaic 3 HAMR platform has already been qualified by all targeted cloud service providers, with management projecting a 50% exabyte crossover to HAMR technology by the second half of calendar 2026.
“We know what’s coming out a year from now,” Mosley said, describing the company’s build-to-order model where customers buy capacity up to four or five quarters in advance. “And we’ve basically gone to the customers and said, ‘Look, if you want to plan this really well… we know what’s coming out. You can buy this stuff up to a certain period.’”
Then came the line that moved markets: “But the demand is significantly higher than that.”
The Ripple Effect: Why Micron, Western Digital, and SanDisk All Fell Together
If Seagate’s problem is capacity, why did completely different companies get dragged down too?
The answer is both simple and a little unsettling: the market prices industries as baskets, not companies.
When a CEO as prominent as Mosley signals that supply can’t keep pace with AI demand, investors don’t just reprice Seagate, they start asking the same question about every memory chip maker. If Seagate can’t scale fast enough, what makes anyone think Micron or Western Digital can?
Compounding the sell-off was a second, simultaneous concern that had nothing to do with Seagate at all. Samsung, which supplies Seagate with chips used in its data storage products, faces potential labor strike risks. If Samsung production slows, Seagate’s already-tight supply chain gets even tighter. Investors connected those dots in real time and hit "sell" across the entire sector.
Broadcom, Nvidia, and Intel all declined as well, not because their businesses are directly threatened, but because the AI trade has become deeply correlated. When one domino falls, the rest follow. That’s the double edge of the AI rally: the same momentum that lifted everyone can reverse just as fast.
"It Would Just Take Too Long", The Logic Behind Mosley’s Blunt Remark
Here’s what’s easy to miss when you’re staring at a 7% daily loss: Mosley’s caution is actually the product of hard-won memory industry wisdom.
Think of it this way. Imagine you’re a farmer who’s lived through both bumper crops and brutal droughts. One season, demand for your wheat goes through the roof. You could borrow millions of dollars, buy more land, and plant every acre you own. But by the time that new land is producing, three to five years later, demand might have cooled, and you’re now sitting on a mountain of unsold wheat with crushing debt payments.
That’s the memory chip business in a nutshell.
Building a semiconductor fabrication facility is a multi-year, multi-billion-dollar commitment. Mosley himself told Bloomberg earlier in May: “The way for us to get to much bigger exabyte production is to jump through these technology trends, not to stop and say, ‘Okay, I’ll build another factory,’ because that’s a long lead time.”
The memory sector has been burned by overcapacity before. When too much supply floods the market, prices collapse. Companies that overbuilt during the last cycle spent years recovering their margins. Mosley, who has lived through those cycles, is essentially saying: I’d rather squeeze more value from the factories I already have than risk building one I might regret.
And to be fair, Seagate is adding tools and equipment within existing facilities. This isn’t a refusal to grow, it’s a refusal to grow inefficiently.
The Technology Bet: Platter Density Over Factory Floors
This is where most articles lose their audience, so let’s make this crystal clear.
Seagate’s entire growth strategy rests on something called HAMR, Heat-Assisted Magnetic Recording, marketed under the Mozaic brand. And no, you don’t need an engineering degree to understand why it matters.
Here’s the metaphor that helps. Picture a bookshelf. You can add more books to your collection in two ways:
Option A: Build more bookshelves (this is building new factories). It takes forever, costs a fortune, and you might end up with empty shelves if your book-buying slows down.
Option B: Figure out how to fit more books on each shelf, thinner pages, smarter stacking. (This is HAMR technology, increasing how much data fits on a single platter.)
Seagate is betting the farm on Option B.
Right now, the company’s Mozaic 3 platform delivers 3 terabytes per platter. The roadmap goes to 4TB, then 5TB, and eventually, through Mozaic 4 and beyond, to 10TB per platter and 100TB drives by 2030.
Why is this so compelling? Because a factory floor that produces 3TB-per-platter drives suddenly produces 67% more storage capacity if you upgrade it to 5TB-per-platter technology, with no new construction, no new buildings, and far less capital risk. Same footprint, dramatically more output.
Mosley’s team has already qualified Mozaic 3 at all planned cloud providers. The company expects 50% of its exabyte output to run on HAMR by the end of 2026. That’s not a science project, that’s a production milestone.
The Supply Chain Reality Most Investors Miss
One of the most underappreciated details in this story is how Seagate actually sells its products now.
The company has transitioned to a build-to-order model. This means customers aren’t buying drives off the shelf, they’re placing orders with four- to five-quarter lead times, essentially reserving production capacity months before they need it.
For Seagate, this is actually a position of strength. It means the company has near-perfect visibility into future revenue. It can plan manufacturing precisely. It avoids building inventory that might sit unsold.
But for investors who want exponential growth right now, it’s frustrating. Build-to-order means you can’t suddenly double output because a hyperscaler placed a massive AI infrastructure order yesterday. The pipe is the pipe.
Here’s the component timeline in plain terms:
- Recording head wafers: 9+ months from order to delivery
- Drive assembly and testing: Another 3 months
- Total pipeline: Roughly 12 months from component order to finished product
Mosley confirmed that demand significantly exceeds current supply, and that unit capacity expansion would only make sense if emerging use cases like Edge AI create demand beyond data centers.
Bull Case vs. Bear Case: What Smart Investors Are Asking After the 7% Drop
If you’re holding STX, or thinking about buying the dip, here’s a framework for sorting through the noise.
The Bull Case
- Sold-out capacity signals pricing power, not weakness. When every drive you can make is already spoken for through 2027, you can command premium prices. Mosley’s team has locked in customers with four-to-five quarter visibility.
- Technology transitions are a legitimate growth lever. Moving from 3TB to 5TB per platter can grow exabyte output by over 60% using the same factory footprint. That’s real growth, it just doesn’t require a ribbon-cutting ceremony.
- The AI storage supercycle is still intact. A January 2026 Morgan Stanley report described the memory market as being in a “generational supply and demand mismatch.” Analysts like Evercore maintain price targets as high as $1,000 for STX.
- Seagate’s Q3 FY2026 results were stellar. Revenue rose 21.5% year-over-year, earnings per share climbed 53% YoY, and gross margins hit 42.2%. The fundamentals haven’t deteriorated, sentiment just shifted.
The Bear Case
- A P/E ratio of ~70x demands perfection. When you’re priced as a high-growth AI beneficiary, any hint of a ceiling on that growth gets punished aggressively.
- Mosley himself told you the ceiling exists. Demand is “significantly higher” than what the company can supply. That’s not a growth problem, it’s a capacity ceiling that won’t budge without major capital investment.
- The Samsung strike risk is real. Even if Seagate’s strategy is sound, a labor disruption at its key chip supplier could throttle production through no fault of its own.
- Macro headwinds are piling up. The 10-year Treasury yield hit 4.63%, its highest since January 2025, tightening the discount rate on high-multiple tech stocks. Oil above $100/barrel and Middle East tensions aren’t helping.
The Honest Middle Ground
The sell-off appears to be more about recalibrating expectations than any fundamental breakdown. Weiss Ratings maintains a “B” (Buy) rating on STX, noting that the company’s Q3 results were strong and that the decline was “driven more by selective institutional repositioning than a broad, panic-driven liquidation.”
AI Memory Demand, Samsung Strikes, and a Supercycle at Risk
Zoom out far enough, and this story stops being about Seagate at all.
We’re in the middle of what analysts have dubbed an AI memory supercycle. Hyperscale cloud providers, Google, Microsoft, Amazon, are spending billions on AI data centers, and every one of those data centers needs staggering amounts of storage. Training large language models consumes storage capacity at an unprecedented rate.
Morgan Stanley’s Joe Moore described the situation as a “generational supply and demand mismatch.” The memory chip shortage is expected to persist throughout 2026 and potentially into 2027.
But here’s the tension at the heart of this moment: the AI buildout requires massive long-term infrastructure bets, and memory chip makers, burned by past boom-bust cycles, are reluctant to make them.
Seagate’s caution isn’t unique. Micron, Western Digital, and SanDisk are all facing the same math: build too much capacity and risk a price war when the cycle turns; build too little and leave revenue on the table.
Mosley’s “take too long” comment is really an honest articulation of that dilemma, and the market’s sharp reaction tells you how much is riding on getting that balance right.
What Seagate’s Caution Tells Us About the AI Infrastructure Boom
Here’s a thought to sit with: maybe the sell-off isn’t just about Seagate. Maybe it’s the first time the market is really grappling with the physical constraints of the AI revolution.
We’ve been treating AI as if it’s made of code and ambition, infinitely scalable. But code runs on hardware. Hardware needs factories. Factories need years, billions of dollars, and specialized equipment that itself has long lead times.
Seagate’s message was simple and, in hindsight, obvious: the supply chain for AI infrastructure wasn’t built for this kind of demand, and it can’t pivot overnight.
The companies that win in this environment won’t be the ones that build the most factories. They’ll be the ones that figure out how to deliver dramatically more value from the assets they already have, squeezing more terabytes from every platter, more revenue from every square foot of factory floor.
That’s exactly the bet Seagate is making. Whether it pays off depends on whether the AI supercycle has legs, and whether the market has patience for a growth story that doesn’t involve shiny new buildings.
Want to stay ahead of the next market-moving moment? Subscribe to our newsletter and get real-time analysis of the AI infrastructure sector delivered straight to your inbox, no fluff, just the signal you need to make smarter investment decisions.