Micron Just Locked In $100 Billion in Sales, and Wall Street Thinks the Boom-Bust Chip Cycle Is Dead

Micron Technology (NASDAQ:MU | MU Price Prediction) had signed 16 long-term customer agreements, 14 of them locking in roughly $100 billion in minimum guaranteed revenue through 2030, and the stock was up 16% because, in her words, “Memory has always been just been boom and then bust. But Micron is essentially telling everyone that’s completely over.”

That claim matters in an industry defined by cycles. DRAM glut, DRAM shortage, repeat. So what did Micron actually do, and why are people who lived through the 2019 and 2023 downcycles willing to underwrite a different ending this time?

What the contracts actually changed

Micron’s fiscal Q3 revenue came in at $41.456 billion, a 17.60% beat on consensus and 345.72% year-over-year growth from the $9.3 billion Micron printed in the prior-year quarter. Non-GAAP EPS landed at $25.11 against a $20.2843 consensus, the seventh consecutive quarter of beating Wall Street.

The margin line is where the cycle thesis wobbles. GAAP gross margin was 84.6%, against 37.7% a year earlier. Software companies have 84% gross margins. Memory companies, historically, have run a fraction of that. Micron is guiding Q4 to around 86% on $50 billion plus or minus $1.0 billion in revenue. As Partsinevelos framed it, “they can determine the pricing right now, which is great for them to keep those margins up.”

CEO Sanjay Mehrotra’s framing in the press release was that “multi-year Strategic Customer Agreements will significantly enhance the durability and predictability of Micron’s strong financial performance.” The hyperscalers building AI infrastructure are willing to pre-commit to HBM and DDR5 capacity through the decade because they cannot afford a repeat of the 2024 shortage. Micron’s Cloud Memory segment alone did $13.769 billion, with Core Data Center another $11.524 billion.

Qualcomm’s parallel bet

Qualcomm (NASDAQ:QCOM) CEO Cristiano Amon pitched the same direction from the other side. Qualcomm raised its non-handset revenue target to $40 billion by 2029, nearly double its prior forecast, with roughly $15 billion from data center. And Partsinevelos noted Qualcomm “named META as its first customer for its new data center CPU, and said it signed two hyperscale deals for custom chips, one in the United States, one Chinese.”

Amon’s answer to how you guarantee any of this was about supply. “I have secure the capacity from the manufacturer as well as memory,” he said, “we’re very confident in the forecast we provided.” That capacity comment validates Micron. Qualcomm’s Q2 handset revenue had already fallen 13% year-over-year to $6.024 billion, with memory supply constraints among Chinese OEMs cited as the cause. The thing squeezing Qualcomm’s phone business is the thing minting Micron’s margins.

Qualcomm shares rose 12% on the data center news, a smaller move than Micron’s, partly because at a $215.827 billion market cap the company is still mostly a handset business until the Dragonfly C1000 actually ships to Meta in 2028.

The execution problem nobody is talking about yet

Micron has to spend to make the $100 billion show up. Q3 capex was $7.826 billion, up 166.37% year-over-year, and the company has signaled capex over $40 billion next year, with roughly $20 billion going to construction and clean rooms. That is a fab-building number. Free cash flow was a robust $18.304 billion in the quarter, so the cash is there, but the rule of memory has always been that whoever builds capacity in the boom regrets it in the bust. The contracts are supposed to be the insurance policy against exactly that.

You can see the market wrestling with the structural-versus-cyclical question elsewhere. UBS tripled Micron’s price target last month, and the stock crossed $1 trillion in market cap alongside SK Hynix. Even retail cannot agree whether this is the end of the cycle or the loudest part of it.

The case for “different this time” rests on contracts whose enforceability has not been tested in a downturn, capacity that does not exist yet, and AI capex commitments from a handful of hyperscalers. Two of those three are denominated in real dollars on signed paper. The third is what you keep an eye on.