Bank of America says DRAM needs to go as high as $568.8B, yet the market smashed SK hynix into a circuit breaker—so who’s gone crazy?

A 325% growth expectation. Circuit breaker. When these two words show up together, one of them is definitely lying.

On July 11, Bank of America dropped a research report: in 2026, global DRAM revenue is expected to grow year over year by 325%, reaching $568.8B. In the same period, NAND is expected to grow 299% to $323.1 billion. Together they total $891.8 billion, up 315% versus 2025. By 2027, the figure is projected to surge to $1.25 trillion.

What does $12.5k mean? It’s higher than the GDP of many countries. This isn’t a cycle—it’s a tsunami.

So what was the market’s response?

On July 13, SK hynix plunged 15.37% in its home market in South Korea, posting the largest single-day decline in history. Samsung Electronics fell more than 10%. The KOSPI index closed down 8.95%, triggering the circuit breaker for the seventh time this year. SK hynix’s total market cap fell below $568.8B, and its share price has cumulatively retreated nearly 40% from its all-time high.

The U.S. stock market crashed in sync: the Philadelphia Semiconductor Index fell 4.78%, SanDisk dropped more than 12%, and SK hynix’s ADR fell more than 9%. The Nasdaq fell by more than 400 points.

In the same industry, one side has a 325% growth forecast, while the other sees a 15% single-day plunge.

Tell me which one is right.

First, get clear on what happened in South Korea.

The direct spark was a report from South Korean brokerage KIS, predicting that SK hynix’s second-quarter operating profit could be about 8% lower than market consensus. The reason: the HBM revenue mix is too high, and HBM is usually locked in with long-term supply agreements that fix prices, so its pace of price growth can’t keep up with the spot market.

Note: KIS’s exact wording was: “This is not deterioration in fundamentals, but a normalization adjustment after incorporating consideration of long-term supply agreements.” The report also maintained a target price of 3.8 million won and a “buy/accumulate” rating.

It was still a bullish report. But because of the phrasing “not meeting expectations,” it ignited one of the largest stampedes in South Korea’s stock market history.

Why? Because Korean retail investors took leverage to the max.

Data from South Korea’s Financial Supervisory Service: the total value of forced liquidations across the whole market on a single day was 344.2 billion won, the largest credit-position stampede this year. Retail investor positions made up 92% of the leveraged book. Across the market, more than 1.2 million leveraged retail trading accounts hit the margin call line; among them, around 320k to 360k accounts were fully forced to close, with principal wiped out to zero—some accounts even ended up owing the brokerage in the negative.

1.2 million people got liquidated. This isn’t a stock-market adjustment—it’s a social event.

A Goldman Sachs trader admitted: “So far, we haven’t heard any truly fundamental catalyst that would be convincing enough to explain the sell-off this time.”

Translation: no fundamental negative news—just leveraged liquidation.

But the other side of the story is what you really should pay attention to.

What are hedge funds doing at the same time?

Goldman Sachs data shows that last week, hedge funds bought the most U.S. semiconductor stocks in nearly three and a half years. Semiconductor stocks now account for 10% of total hedge fund exposure—double the same period last year.

They’re疯狂疯狂ly buying.

SK hynix’s ADR listed on Nasdaq last Friday with an issue price of $149, surging 12.8% on its first day. On Monday, even though it fell 9% due to the Korea drag, it traded in a market with better liquidity, fewer leveraged positions, and institutional pricing.

Two markets, two pools of capital, two different pricing logics:

Korea market: retail investors add leverage to bet on storage, 1.2 million get liquidated, and they’re forced to cut losses.
U.S. market: hedge funds add aggressively against the tide, betting that the selling is over.

Do you believe retail panic—or do you believe institutional ammunition?

A 325% growth expectation—does it really only deserve a circuit breaker?

Bank of America’s logic: this round of growth is driven by price rebounds, not just by shipment growth. It expects the average DRAM price in 2026 to rise 249% year over year, and NAND to rise 238%. A triple force: a surge in demand for AI servers, HBM continuing to ramp up, and supply constraints.

Last Friday, SK hynix’s CEO said: “The shortage situation for memory chips may continue beyond 2030.”

In a seller’s market, supply-demand gaps that can’t be filled on a year-by-year basis, and a doubling-style rise in ASP—tell me, is this what people call “cycle topping”?

MRM Research analyst was even more direct: SK hynix’s stock currently shows “deep oversold” characteristics—“it’s not impossible for it to drop another week, but we believe this is a good opportunity to add.”

Finally, let’s talk about what’s truly important for the crypto space.

DRAM and NAND entering a super cycle means compute costs will stay high.

HBM prices are constrained by long-term agreements—so they’re rising too. And the spot market is even more so. Inflation in AI hardware has already been built into all assumptions behind discounted cash flow: compute costs rise, and capital expenditure intensity increases. The valuation of the tech sector is no longer just a “growth story,” but includes a scarcity premium on resources.

What does this mean for crypto infrastructure?

ZK proofs, AI agents, Layer2 sequencers—every track that depends on compute has its cost structure being rewritten. For projects in the storage track, if they’re still using a “Moore’s Law price declines” logic to build financial models, it’s time to recalculate.

The DRAM super cycle isn’t just something the semiconductor industry deals with. It’s the underlying inflation across the entire digital world.

“Panic is temporary; cycles are eternal. Don’t let a circuit breaker scare you to death—look behind the circuit breaker. The industry is going through a major boom.”

Early on July 14, South Korea’s KOSPI already rebounded and turned positive. Samsung Electronics rose more than 5%, and SK hynix rose more than 3%. Panic lasted only one night.

A 325% growth forecast, a 15% plunge—what do you believe? In your crypto holdings, is there a token that’s benefiting from “compute inflation”? #PreIPOs第二期OpenAI认购 #百万充值补贴 #沃什听证会撞上CPI $BTC $SKHY $SAMSUNG