Micron CEO is doubling down on a cycle-free future

Micron CEO is doubling down on a cycle-free future

Tobi Opeyemi Amure

Mon, August 24, 2026 at 1:07 AM GMT+3 7 min read

• MU -0.77%

Every industry eventually produces an executive willing to say the thing out loud. The old rules do not apply anymore. The machine has changed.

Sometimes that turns out to be true. Often it marks the top.

Memory chips are the purest cycle left in modern manufacturing. Demand climbs, prices follow, every producer races to add capacity, and the capacity lands about two years later, all at once, into a market that no longer needs it. Prices break. Profits vanish. Whoever is still standing cuts spending and waits.

Micron Technology (MU) has run that loop since 1978, and it has the receipts. In fiscal 2022, the company earned $8.7 billion. In fiscal 2023, it lost $5.83 billion as revenue nearly halved to $15.5 billion, according to Micron's annual filings.

That is three years ago, not three decades.

Which is what makes this week worth sitting with. Chief Executive Sanjay Mehrotra is now arguing, with $10 billion behind the argument, that the loop has been cut for good.

"So, the value of memory, that equation has totally changed," Mehrotra told CNBC on Aug. 20, speaking from a fab construction site outside Boise, Idaho.

What Micron is actually buying with $10 billion

The thing Micron announced is not a factory. It is a research institution.

The company unveiled Micron Research Labs on Aug. 20, headquartered in Boise and backed by a planned $10 billion investment over the next decade, according to Micron. Ground breaks in calendar 2027. The work covers memory technologies, compute architectures, packaging and future semiconductor manufacturing, and the stated horizon stretches past 10 years.

Related: SanDisk sends strong signal to Micron investors, BofA says

That is the kind of spending "that sits upstream of every product we build," said Scott DeBoer, Micron's chief technology and products officer.

It sits on top of the more than $250 billion Micron has separately committed to U.S. manufacturing and research through 2035, according to the company.

None of it ships a chip this decade. That is the entire point, and it is also the problem.

Why the memory cycle has always come back

The cycle is not a failure of nerve. It is arithmetic, and the arithmetic has not changed.

A fab takes three to four years from concrete to wafers. Demand signals move in quarters. By the time the supply answer arrives, the question has changed. Every memory boom in the past 40 years has ended the same way, and none of them ended because producers were stupid. They ended because the lag is structural.

Wall Street has not forgotten. Micron is a business where the DRAM and NAND markets are "highly cyclical," according to Morningstar, whose analysts still decline to award the company an economic moat despite its scale.

More Artificial Intelligence:

Mehrotra's counterargument is that AI changed the buyer, and that the change is structural. Memory now gets designed alongside the processor it will sit next to, which pulls Micron into a customer's roadmap years ahead of the order. Data center customers currently want about 50% more supply than Micron can commit to, he told CNBC.

He is not alone in that read. Jim Cramer argued that AI has rewritten the rules for memory stocks and that Micron can double from here, TheStreet reported. Bank of America has made a version of the same case, pointing to the industry's shift toward multi-year supply agreements as a reason this upcycle should behave differently, which was also highlighted by TheStreet.

The bank was careful about the wording, though. Those contracts do not prevent a downturn. They shape how one arrives.

Micron has been converting that thesis into paper. The company disclosed 16 five-year strategic customer agreements alongside its June earnings report, and Mehrotra said more have been signed since.

Those contracts, not the research campus, are the actual cycle-proofing. The lab is the flag planted on top of them.

The numbers Micron does not put in a press release

When I lined up Micron's fiscal 2023 filings against this week's announcement, the same tension showed up in both directions.

Here is the last full turn of the wheel, and where the company sits now.

• Fiscal 2022 net income of $8.7 billion, one of the best years in company history, according to Micron's annual results.

• Fiscal 2023 net loss of $5.83 billion on revenue of $15.5 billion, down from $30.8 billion, according to Micron's annual results.

• Fiscal 2023 gross margin of negative 7.3%, meaning chips sold below the cost of making them, according to FactSet data cited by CNBC.

• Capital spending cut to $7.7 billion in fiscal 2023 from $12.1 billion the prior year, according to CNBC.

• Trailing 12-month net income of roughly $50.5 billion as of mid-August 2026, according to The Motley Fool.

The fourth bullet is the one that matters for a research lab. Micron kept investing through the last downturn, but the CEO has said plainly that the spending got cut back hard from the year before. When margins went negative, the company had no choice.

Mehrotra made a related point on CNBC in June, arguing that years of customers squeezing on price left the whole industry underinvested right as AI demand arrived. That is a candid admission, and it cuts both ways. If price pressure once dictated Micron's research budget, price pressure can dictate it again.

What would actually prove the cycle is broken

My read is that the $10 billion figure is the least interesting number in this story.

A decade-long research commitment announced during the best quarter in company history costs Micron almost nothing today. It is funded out of a trailing net income figure that dwarfs the entire pledge. The test is not whether Micron writes the check now. It is whether Micron writes it in the fiscal year DRAM prices roll over and gross margin goes red again.

That test is not hypothetical, and it is not far off. Micron shares closed at $974.33 on Aug. 20, up about 4%, and still sit roughly 22% below the record close they set in late June, according to market data. The market has already started pricing cycle risk into a stock whose CEO says the cycle is over.

Investors got a preview in June, when Micron gave back weeks of gains in two sessions on nothing more than a broad chip selloff. Nothing about the demand story changed those two days. The stock moved anyway, which tells you what the market still believes underneath the narrative.

For anyone outside the trade, the stakes are more immediate than a chart. Memory scarcity has been pushing up prices on phones, laptops and consumer electronics, a squeeze Mehrotra himself traced back to industry underinvestment. The same shortage that made Micron one of the best-performing AI trades of 2026 is showing up on your next hardware invoice.

Watch the next earnings report for the first tell. Micron's fiscal fourth-quarter results are due in late September, according to market data, and the line I would go to first is not revenue. It is research and development expense, and whether management is willing to put a floor under it in writing.

Cycles do not die because an executive announces they have. They die when a company keeps spending through the year that would normally force it to stop.

Related: Micron CEO gives investors $10 billion reason to listen

This story was originally published by TheStreet on Aug 23, 2026, where it first appeared in the Technology section. Add TheStreet as a Preferred Source by clicking here.