Let’s get it straight: Micron’s Q4 earnings report lands after the US stock market closes tonight (at around 1:00 a.m. Beijing time on October 1). So the real question right now isn’t whether the “numbers” are good—because the figures have already been guided and essentially locked in (US$50 billion in revenue, 86% gross margin, and Micron has beaten expectations in each of the past seven quarters). Instead, can it validate the words “super cycle”?
Tonight’s earnings report only tests three things—what matters more than the numbers themselves:
Q1 FY2027 guidance: If the January quarter—a traditional off-season—still grows quarter over quarter, then cycle believers can shut up;
Can the gross margin hold at 86%? Proof of structural pricing power if it doesn’t fall;
Wording for capacity and long-term contract coverage in 2027–2028: This is the dividing line between “structural” and “cyclical.”
So how long can the super cycle last? I lay out the answers from three major banks, and you’ll see that disagreement itself is the answer.
BofA (most conservative): In 2026, the DRAM supply-demand tightness is only 79%. Cloud vendors have already signed contracts with prices in Q1 2027 higher than Q4 2026. But in 2028, the average price falls 10% from its peak—an “orderly soft landing.”
Citi (most optimistic): In 2027, DRAM demand grows 30.2% versus supply growth of only 18.8%, leaving a gap of 8.7%. In 2028, the gap widens further to 9.7%; shortages could persist through 2031.
The middle camp (SK hynix itself): supply-demand balance won’t happen until at least 2030; the cleanrooms in the new plants won’t be operational until as early as Dec 2028.
The special thing about this cycle is that the “money-printing machine” is welded in place by contracts.
In the past, storage cycles were like a pendulum: price hikes → mad expansion → oversupply → a brutal crash. Even in 2023, Micron was still losing money. This time it’s different: major customers sign take-or-pay agreements—pay even if they don’t take delivery—about $100 billion in minimum underwriting, coverage extending to 2030, plus $22 billion in customer prepayments.
Even if prices peak and fall in 2028, the profit curve will most likely be a “downhill slope” rather than a “free fall.” BofA raised the 2030 memory-market TAM from $1.8 trillion to $2.0 trillion, with a 21% CAGR from 2027–2030. This is the theoretical basis for the cycle stocks being forcibly re-rated as growth stocks—PE multiple repair from 6–8x up to 20x.
But you also have to see the three cracks clearly:
The first derivative of the rate of growth has already peaked: DRAM contract price Q1 QoQ +90–95% → Q2 +58–63% → Q3 only +13–18%; Micron’s revenue QoQ growth rate slowed from +74% to the guided +21%. Slowing growth doesn’t necessarily mean the cycle is over, but the steepest part is already behind us.
The consumer end has hit its limit: mobile phone and PC buyers refuse to follow price increases. Demand destruction is the classic way price-hike cycles die.
Capacity concentrated release in 2028: the three companies simultaneously pour money in (SK hynix +$38.3 billion, Micron +$27 billion, Samsung doubling). The gap could be filled all at once in 2028–2029. Citi’s optimistic 2031 scenario hinges on AI demand never stopping.
Price momentum can still burn into the first half of 2027. High-margin levels can be supported by long-term contracts into 2028, and in 2028–2029 we’ll see how real the AI demand is. But note one brutal fact: the market only believes a two-year story. Micron’s forward PE is only 6–7x, while the S&P is 21x. Wall Street uses valuation to cast a vote: “I’ve seen too many super-cycle corpses.” Analysts’ average target price is $1,303–$1,523 (highest $2,000) versus a current price of about $1,000. The entire bull-bear split comes down to the phrase “this time is different.”
Your landing point:
No need to bet on direction tonight—tomorrow morning, watch two lines of text: whether the Q1 guide QoQ is positive, and whether the gross margin guide is ≥86%. If both are green, the super-cycle thesis is crowned; if one is red, cut back half the position first.
For holdings in the storage supply chain: set mid-2027 as the first de-risking observation line (the intersection point where new capacity ramps up and the price-increase slope starts to slow).
To make money from “valuation repair”: if the earnings validate the structural thesis, the upside potential for PE to repair from 6x to 15x is bigger than the elasticity of chasing price spikes.
Risk warning: Once the super-cycle narrative is disproven (even slightly—say, guidance misses by a small margin), the “double hit” to valuation and earnings will happen at the same time. The 11% options implied volatility also shows the market isn’t fully confident; storage capacity ramp-ups in China, slower capex from hyperscale cloud vendors, labor-management disputes in Taiwan—any one of these could end the cycle early. Storage is a strongly cyclical industry. The losses in 2023 are only three years away. This piece is a pre-mortem scenario; remember to come back after the earnings are released to see how it turns out.
$MU