SanDisk, Micron & Seagate Are Down Around 10% in a Week; Crash or Buying Opportunity?

Memory and HDD stocks gave investors one of the most extraordinary runs in recent market history with SanDisk up nearly 3,000% in a year, Seagate up 225% in 2025 alone, Micron up 240% across 2025. And then, in the last week, all three erased about 10% of their market value. SanDisk slid 11.6%, Micron dropped 7.2%, and Seagate fell 8%. 

The question every investor is now asking: is the AI memory supercycle cracking, or is this simply the market doing what it always does, finding reasons to pause before running again?

Let's break down what actually sparked the sell-off, what Seagate's CEO said that sent tremors across the sector, why demand concerns post-2028 are emerging, where analysts still see the stock going, and critically what the premarket recovery this morning is really telling us.

Why Memory Stocks Suddenly Crashed? Seagate CEO That Spooked the Whole Street

On Monday, May 18, Seagate Technology CEO Dave Mosley stepped up to the mic at the JPMorgan Global Technology, Media and Communications Conference and, when asked how the company planned to ramp up factory capacity to meet surging AI demand, said something that the market didn't love:

"If we took the teams off and started building new factories or bringing up new machines, that would just take too long. You would end up with more capacity, but then you'd slow the rate of growth on that technology."

He went further, adding that demand is "significantly higher" than the four-to-five quarter visibility window Seagate can currently guarantee. In plain English: we can't build fast enough to keep up, and we're not even going to try.

Now here's the paradox that the market got tangled in. If supply can't meet demand, prices should stay elevated, which is actually good for Seagate's margins. But investors read it differently: if capacity growth is capped, revenue growth has a ceiling. And for stocks trading at multi-year highs after multi-hundred-percent rallies, any ceiling is terrifying.

The immediate damage:

Source: Yahoo Finance

A second worry piled on: Seagate uses Samsung chips in its hard drives, and there were simultaneous concerns around a potential Samsung labour strike, which could further tighten an already constrained supply chain.

Context You Need: These Stocks Had Already Run an Extraordinary Race

Before we analyse the fall, let's appreciate what came before it. Think of it like this: if a Rajdhani Express has been doing 160 km/h for hours and suddenly slows to 140, is it broken or is it just adjusting speed on a curve?

Here's the full scoreboard of how these three stocks have performed in recent times:

Source: Google Finance

*SanDisk was spun off from Western Digital and listed in Feb 2025, making its 1-year return extraordinary due to low base + NAND pricing supercycle.

These were not slow-and-steady compounder returns. These were some of the most explosive stock moves in S&P 500 history. In that context, a 10% correction is less a crash and more the market exhaling.

Is the AI Memory Supercycle Breaking?

The bull case for memory stocks has always rested on two pillars: 

1. AI-driven demand is structural and multi-year; 
2. Supply can't keep up. 

Both remain true right now. But a third variable is quietly entering the room: what happens when new fabs finally come online?

IDC estimates 2026 DRAM supply growth at only 16% year-on-year and NAND supply growth at 17%, both well below the 20-30% historical norms. Meanwhile, TrendForce projects the total memory market to reach $551.6 billion in 2026 and surge to $842.7 billion in 2027.

So demand is racing ahead and supply is lagging. That's the tailwind. But here's the cautionary note: getting back to normal pricing and availability looks more like 2028 to 2029, and that timeline assumes AI infrastructure demand grows at current forecast levels rather than accelerating further.

The flip side? If AI hype goes down unexpectedly, due to regulation or lack of immediate profitability, demand projections could crash. If memory makers were to over-invest, they could face severe excess capacity by 2028. This is precisely why many have held back on greenlighting new capacity lines.

Think of it like India's solar capacity boom. Between 2020 and 2024, demand raced ahead, prices were high, and everyone wanted solar panels. Then new manufacturing scaled up and, in some segments, oversupply entered the picture. The AI memory story is tracing a similar arc, just on a compressed timeline and with vastly higher stakes.

Analysts projected that the memory price rally is poised to extend past 2028. Pricing data from Q4 2025 reinforced that outlook, with DDR5 chip prices rising from $6.84 in September 2025 to $27.20 in December.

Our read: the post-2028 risk is real but not imminent. What investors are pricing in right now is a fear that the cycle will peak earlier than models suggest and Seagate's CEO accidentally gave them a reason to price that fear in.

Why Analysts Still Remain Bullish on Memory Stocks

Despite the sell-off, Wall Street's collective view on these three stocks has not turned bearish. In fact, the analyst community is more constructive on SanDisk and Seagate than the week's action would suggest.

Sources: Benzinga, TheStreet, TickerNerd, ChartMill, Barchart, S&P Global Market Intelligence

One nuance worth pointing out: for Seagate, the median analyst target of $771 actually sits below the  price which was around $834 before this week's correction, suggesting that the Street broadly already saw the stock as having gotten ahead of itself, even before the sell-off. Citigroup actually lowered its Micron target from $510 to $425 earlier this month, a notable bear-side call amid the broader bullishness.

For SanDisk, the story is more dramatic. Analysts are revising earnings estimates sharply upward; Morgan Stanley's Joseph Moore models SanDisk earning $127.92 per share in calendar 2026, 65% above Wall Street consensus. For 2027, his estimate of $149.68 sits 42% above the Street's $105.03. When a top-tier firm is this far above consensus on a name, it's either a very high conviction call or a sign of deep uncertainty about the cycle's durability, and in this case, it's both.

The Pre-market Recovery: Is the Correction Already Over?

On the morning of May 20, 2026, all three stocks have bounced in pre-market trading:

Source: Google Finance

This is meaningful, but don't mistake a bounce for a bottom signal.

What's driving the recovery? Two factors: 

• First, Citi upgraded its SanDisk price target to $2,025 on May 19, right in the middle of the sell-off, a notable vote of confidence. Citi analysts led by Asiya Merchant said hyperscalers are accelerating purchases of enterprise solid-state drives as they expand generative AI training and inference infrastructure. The firm expects NAND average selling prices to surge more than 186% year over year in 2026, with enterprise SSD pricing rising even faster as supply remains constrained.
• Second, the broader semiconductor index (SOX) began stabilising, with institutional buyers likely stepping in on dips in names they already had conviction on.

But here's the honest caveat: a premarket bounce after a sell-off does not confirm that the correction is over. Memory stocks are highly cyclical by nature. The market can re-test lows if macroeconomic data deteriorates, if China demand softens, or if any hyperscaler signals a pause in capex. Seagate has already flagged it will speak at TD Cowen's tech conference on May 27 and Bank of America's global tech conference on June 2, those events will matter.

Our Analytical Take: The 3 Things That Actually Matter Now

Here is the core tension that every investor needs to sit with:

1. The Seagate CEO's comment was NOT bearish on demand, it was bearish on supply expansion.

Mosley wasn't saying demand is slowing. He was saying building new capacity takes too long to be worthwhile. That means existing capacity stays tight, pricing stays elevated, and margins stay high. The market initially read this as bad news; the more correct reading is that it is constrained upside, not downside risk. There's a difference.

2. The cycle is real, but the clock is ticking toward 2028.

Market research firm IDC estimates that global data generation could increase by more than fivefold between 2020 and 2028, driven by growing usage of AI applications. As a result, the need for data centre storage could more than double between 2024 and 2028. That's the secular growth story. But the risk is that new capacity from SK Hynix, Samsung, and Micron's planned fab expansions all come online between 2027 and 2029, potentially creating an oversupply window when AI capital spending inevitably normalises.

3. These are still cyclical stocks dressed up in an AI narrative.

This is the differentiating point that most investors miss. Micron, Seagate, and SanDisk are extraordinary beneficiaries of a once-in-a-decade demand surge. But they are not protected from the memory industry's oldest pattern: boom leads to overinvestment leads to glut. The AI story has extended and arguably deepened this cycle, but it has not abolished the cycle. Since the memory sector is highly cyclical, a slowdown in order growth could lead to oversupply, pressuring prices and profitability.

The investor who understands this distinction is in a very different position than one who sees these as "buy and forget" AI infrastructure plays.

Is This Correction a Buying Opportunity?

Management's comments on the latest earnings call suggest Seagate could continue to raise prices on new contracts to capitalize on the shortage of HDDs and SSDs, which could persist until the end of the decade as AI server shipments increase.

Analysts estimate AI data centres could consume approximately 70% of high-end DRAM in 2026, a dramatic inversion from prior cycles, with DRAM contract prices having jumped 50%+ QoQ entering 2026.

The structural case for memory demand through 2027 is intact. What's changed this week is investor confidence in the pace of upside, not the direction. And when sentiment corrects on high-momentum stocks that have run up 500-3,400%, a 10% drawdown is not a crisis. It is, frankly, normal.

• For investors with a 12-18 month view: The demand story remains in place. The correction provides an entry point that was not available last week. The risk to watch is post-2028 capacity normalisation and any macro signal that hyperscaler capex is being pulled forward and borrowed from future quarters.
• For short-term traders: This is a volatile sector. The premarket bounce is encouraging. The next real test is what Seagate says at JPMorgan and Bank of America conferences in late May and early June. Watch those closely.

The 2028 window is the one to watch. Not because the story ends there, but because that's when new capacity, normalising AI capex cycles, and potential pricing pressure could intersect. Between now and then, the fundamentals remain compelling.