The Rhyme: Micron Bets $250B on AI & Its 2000 Echo

"History doesn't repeat… but it rhymes." — Mark Twain

Micron Technology raised its U.S. chip investment pledge to more than $250 billion through 2035, up from the $200 billion it promised just months ago. Its planned New York facility will be the largest semiconductor manufacturing site in U.S. history, backed by the largest private investment in New York state history. The stock jumped 7% to $1,021 on the announcement, dragging the broader chip sector higher after weeks of selling that had taken the semiconductor index down 16% from its June peak. Micron has been the trade of 2026 — up 241% this year, past a trillion-dollar market cap in May, and now betting its entire future on a demand curve that only works if AI spending never slows down.

MU +7% to $1,021 | $250B committed through 2035 | SOX −16% from June peak

The last time a single company made a bet this big on building capacity for a technology everyone agreed would grow forever, the company was headquartered in a small town in upstate New York. It made glass.

Corning Incorporated reported its second-quarter earnings that morning from One Riverfront Plaza, a low-slung brick building on the banks of the Chemung River in a town of twelve thousand people. The numbers were staggering. Earnings per share had jumped 80% from a year earlier. Revenue hit $1.78 billion, up 57%. Demand for optical fiber — the hair-thin glass threads that carried the internet's data traffic under oceans and across continents — was growing at 40% a year and showing no sign of letting up.

CEO Roger Ackerman stepped to the microphone and called it a "blockbuster quarter." He raised the full-year earnings outlook by 60% and told the analysts crowded onto the call that the company was investing in capacity as fast as the concrete could cure. Corning had already committed more than a billion dollars to new fiber optic plants in North Carolina and abroad. It had spent $2.1 billion acquiring NetOptix, a maker of optical coatings, and billions more on Oak Industries before that. The company's revenue had leaped from $4.7 billion in 1999 and was on pace to hit $7.1 billion by December. Every quarter was bigger than the last, and Ackerman saw no ceiling.

The logic was simple and universal on Wall Street. Internet traffic was doubling every hundred days — or so the Commerce Department had claimed. The Telecommunications Act of 1996 had thrown open the doors for new carriers, and every one of them needed fiber in the ground. Corning made the best fiber in the world. Its stock had climbed from $9 in August 1998 to nearly $100, and the analysts were still raising their price targets. At investor conferences that summer, portfolio managers talked about Corning the way they talk about Nvidia today — as if the laws of supply and demand had been permanently rewritten.

What nobody in that small town on the Chemung River could see was that the carriers placing those massive orders were financing them with junk bonds and borrowed optimism. The claim about traffic doubling every hundred days was off by a factor of ten. The real number was closer to once a year. Within eighteen months, carriers would stop building, orders would evaporate, and 95% of the long-haul fiber Corning was racing to produce would sit dark underground, connecting nothing to nowhere.

The stock peaked near $100 in early September. By late 2002 it traded around $1. The technology wasn't wrong — fiber optic cable did become the backbone of the modern internet. But the buildout ran five years ahead of the demand curve, and Corning was the company left holding the bill for factories it no longer needed.

Two infrastructure makers, twenty-six years apart, making the biggest bets in their industries at the exact moment everyone agrees the demand will never stop.

The rhyme is tight, but the differences matter. Here is where the pattern breaks.

After that July 17 earnings call, Corning kept building. The stock touched its high near $100 in early September. Then the telecom carriers started quietly pushing back delivery schedules. Not canceling — just delaying. A quarter here, a quarter there. By the spring of 2001, the delays had turned into cancellations. Roger Ackerman retired as CEO in January 2001, handing the job to John Loose, who inherited a company that was about to lose nearly everything it had spent two years building.

The layoffs came in waves. Corning cut thousands of jobs in the fall of 2001. Then thousands more. By the time it was done, the company had eliminated over twelve thousand positions — roughly a third of its workforce — and shuttered fiber plants on three continents. Revenue fell from $7.1 billion in 2000 to $3.2 billion by 2002. The billion-dollar factories the company had raced to build sat empty and dark, much like the fiber they were designed to produce.

The industry's spending numbers told the full story. Telecom carriers had poured roughly $120 billion into infrastructure in 2000. By 2001 that number fell to roughly $100 billion. By 2002 it dropped under $50 billion. WorldCom filed for bankruptcy in July 2002, the largest filing in American history at the time. Global Crossing had gone down six months earlier. The customers Corning had built for simply ceased to exist.

The investors who came through it whole were the ones watching one thing: the gap between what a company was spending and what it was earning. When Corning's combined capital spending and acquisitions blew past a third of its revenue in 2000, the math only worked if demand kept growing at the most optimistic projection for a decade running. Micron's capex this fiscal year will exceed $25 billion, and next year's is guided to jump by another $10 billion or more. The direction is clear. The only question is whether anyone is watching the gap.

Corning's stock peaked two months after its CEO called demand "unprecedented" and pledged to build as fast as humanly possible. Not because the internet wasn't coming — it was. But because the buildout got five years ahead of the demand curve, and the market doesn't grade on a pass/fail timeline. When a trillion-dollar chip company pledges a quarter-trillion dollars because AI demand is "structural," the only question every investor should be asking is: structural on what timeline?

The June CPI report drops Tuesday. Inflation is running at 4.2% year-over-year, the hottest since April 2023, and the Fed's next decision comes July 29. If the print lands above consensus, the parallels to June 1980 — when the Fed was forced to reverse a premature rate cut and re-tighten straight into a recession — start getting harder to ignore.