Memory prices up over 500%. Does that still count as good news? A cycle view, and how I trade it with sell puts

Written Thursday, 24 Sept (SGT).

First, is the 500% number real?

Mostly, yes. It depends on which chip and which price you look at.

For the newest standard memory, DDR5 is up nearly 500% year on year. The spot market is even more extreme: DRAM spot prices have surged nearly 700% over the past year, according to a Bloomberg report from July.

Contract prices, which is what big buyers actually pay, moved in huge quarterly steps. TrendForce measured conventional DRAM contract prices up 93% to 98% quarter over quarter in Q1 2026, with many DDR5 SKUs up 3.5 to 4 times and DDR4 kits roughly doubling. Even after all of that, prices were still rising in August. PC DRAM averaged a record $25 that month, up more than 4% from July.

The most extreme example is AI memory. A 36GB HBM3E module sells for $2,100 on the spot market. That is four to five times the long-term contract price.

So "over 500%" isn't hype. The real question is who it's good news for.

Why this is happening: a structural shortage

This is not the usual cycle where everyone overbuilds and prices crash. Three companies control over 95% of global DRAM production, and they have been shifting capacity toward high-bandwidth memory for AI accelerators. That leaves standard DRAM and NAND in very short supply.

AI demand is also growing much faster than supply. A typical AI server uses roughly eight times more memory than a traditional server, and AI server memory spending is on track to grow from $35 to $40 billion in 2025 to $175 to $190 billion by 2027. On the supply side, the limit is physical. You can't build clean room space quickly.

Long-term contracts make the squeeze worse for everyone else. Samsung's memory division has reportedly committed about 70% of its production capacity through 2031 to long-term agreements. That leaves much less for the spot market.

The winners: memory makers are earning at record levels

Micron: In fiscal Q3, DRAM revenue went from $7.1 billion a year ago to $31.3 billion, and NAND revenue climbed from $2.2 billion to $9.9 billion. Almost all of that came from higher prices, not higher volume. DRAM volumes rose only in the low single digits quarter over quarter, while DRAM prices rose in the low 60% range. For the quarter it reports on September 30, analysts expect earnings of $31.26 per share, up from $3.03 a year earlier, on revenue of $50.78 billion versus $11.31 billion.

SK Hynix: Revenue exceeded 50 trillion won for the first time, with an operating margin of up to 72%.

The whole industry: Memory industry revenue is forecast to surge from $220.1 billion in 2025 to $837.3 billion in 2026.

For memory makers, this is the best period in their history.

The payers: everyone who buys memory

This is the part the headline leaves out.

Phones: In flagship phones, DRAM and NAND together made up about 40% of the component cost in Q2 2026. DRAM overtook the processor to become the most expensive single part. IDC expects global smartphone shipments to fall 16.7% in 2026, the largest annual decline on record, while the average selling price rises 27.6% to $581.

Apple: Even Apple is paying. Its CFO said rising memory costs were the main reason for sequential gross margin pressure in the June quarter.

Nvidia: The same pressure is in its guidance. Gross margins are expected to bottom at 71% to 72% in Q4 of fiscal 2027, as rising memory costs weigh on profitability.

So a 500% memory price rise is great news for about three companies and a cost increase for almost everyone else in tech hardware.

Is this the peak?

This is the most important question, and the honest answer is that nobody knows yet. A useful test is to watch three signals that usually show up before a memory cycle peaks: the spot premium shrinking, unit prices flattening, and export volumes recovering. None of those is happening yet.

The bull camp has strong support. Bajarin's analysis argues that three preconditions for a cycle peak (trough inventory, peak year-over-year pricing, and a turn in the stock price) remain unmet, and that memory industry profits are not expected to peak before Q4 2027. SK Hynix itself has said the DRAM supply-demand imbalance will get worse in 2027.

But there are clear risks to keep in mind:

New supply from China. Chinese producers like CXMT and YMTC pose a structural supply threat, with the meaningful impact expected in 2027.

Competition in AI memory. SK Hynix's share of HBM revenue fell from 58% to 50% in Q2, while Samsung's rose from 21% to 33%.

Demand destruction. When phone shipments fall at record rates, high prices are already cutting demand. That's how every memory cycle has eventually turned.

Contracts limit Micron's upside. Micron's long-term supply agreements cap how much of the price rise it can capture.

My take: the good news is real, but it's the kind of good news that sits late in a cycle, not early.

How I use this in my sell-put strategy

Memory stocks are not on my core list. But this topic teaches the most important lesson in my process, so here's exactly how I think about it.

1. Cyclical stocks look cheapest near the top.

The lowest P/E on a cyclical stock often comes when earnings are at their peak. Micron trades at a forward P/E of around 6, with $18 billion in customer cash deposits and $100 billion in contracted minimum revenue. That looks like a bargain. But if prices peak, those earnings drop fast and the "cheap" multiple disappears. So I don't treat a low P/E on a memory stock as a safety margin for a short put. My strike has to be justified by price and volatility, not by the valuation.

2. The earnings calendar decides first.

Micron reports on September 30. My rule is no new position within 7 calendar days of earnings, so MU is blocked from September 23. It doesn't matter how strong the story is. Earnings on a stock like this can gap it 10% or more in either direction.

3. Check the size before you even look at the chart.

Micron shares recently traded around $1,077. One cash-secured put at a $1,000 strike ties up $100,000. My cap for non-core names is 2% of the account per position, which means one contract per $5 million of account value. SanDisk is even more extreme. For most retail accounts, a single memory-stock put is a much bigger bet than it looks.

4. After earnings, let the volatility reset.

Implied volatility usually collapses the morning after a report. That often makes my IV/HV gate (implied volatility at least 1.2 times historical) fail, and I'm not selling cheap premium. The setup I'm waiting for is a "sell the news" drop after the report: the stock falls back toward its moving averages, fear pushes volatility up again, and the IV Rank (40% or more) and IV Percentile (45% or more) gates line up.

5. Where to put the strike.

I take the lower of two anchors: the one standard deviation expected move below the price, or the nearest key moving average × 0.98. Then I check the delta is between 0.15 and 0.25. After a big run, those two anchors are often far apart. A post-earnings drop is what brings them back together.

6. Memory is one trade, not three.

Micron, SanDisk, and the memory suppliers behind Nvidia all move on the same price chart. If I hold puts on more than one of them, I count them as a single position for risk purposes. On top of that:

a 50% take-profit order goes in on fill,

stops are 2.5 times the premium for non-core names,

one roll at most.

My verdict

Memory up 500% is fantastic news for memory makers, costly news for device makers and AI chip buyers, and a warning for anyone buying late. The shortage is real and may last into 2027. But in a cyclical industry, the best headlines usually come late in the cycle.

For a put seller, this is a time to be patient, not greedy. Let September 30 pass, let volatility reset, and only sell a put when the gates line up and the size fits the account. The story being true doesn't make every entry safe.

Do you think this cycle is different, or is memory just being memory again? Let me know in the comments.

This is my personal trading framework, shared for education. It is not investment advice. Options carry significant risk, including assignment and losses larger than the premium collected. Always verify live chain data and confirmed earnings dates before trading.

$MU $SNDK $NVDA $AAPL