Micron Bulls vs Bears: Is the Memory Boom-Bust Cycle Really Over?

The Micron debate comes down to one question

The fight the market is having over Micron boils down to a single line: is this time different? Memory has been the textbook cyclical industry for decades, rising and crashing on repeat. The bull case says AI broke that cycle; the bear case says the cycle never dies.

I respect both arguments, so instead of picking a side up front, I'm going to put three of each head to head.

The bull case: AI broke the cycle

1. AI is starving for memory like never before. Every time a chatbot answers, every time a data center runs an AI workload, it needs enormous memory — especially HBM (high-bandwidth memory). The AI chips Nvidia builds consume three to four times more wafers than regular memory. Even running flat out, supply can't keep pace, and when it can't, prices rise. Micron's pricing power right now is the strongest in its history.

2. The boom-bust cycle is supposedly done. Micron locked in roughly $100 billion of revenue through 16 long-term agreements. Customers put down deposits, and price floors are written into the contracts — even if the market cools, Micron collects a minimum. AI companies pre-committing years of volume means Micron has arguably become a predictable secular grower. That structural shift lines up with the view in the 2027 memory supply monopoly.

3. For an AI stock, it's cheap. This one surprises people. Despite all of it, Micron trades at roughly 9–14x forward earnings. Nvidia sits at 19–22x, Broadcom near 32x. Bulls call Micron a discount ticket into the AI infrastructure trade. Bank of America has the street's highest target and still calls it deeply undervalued.

The bear case: the cycle never dies

1. This industry has always crashed, and it will again. The pattern never changes: prices spike, everyone makes money, everyone races to build fabs, supply floods in, prices collapse, losses follow. Bears expect new fabs coming online in 2027 to add supply and evaporate Micron's pricing power. They're not saying AI is fake — they're saying this industry has never escaped its own nature.

2. Priced for perfection is a dangerous place to live. An 86% gross margin is astonishing, but a price that assumes everything goes right reacts far more violently to any bad news. If margins slip, a big customer pulls back, or consumer demand softens and NAND prices fall, the stock doesn't dip — it drops hard. At this valuation there's almost no room for error.

3. Will AI need this expensive memory forever? This is the least-discussed bear case. Training the largest models needs vast HBM today. But AI keeps getting more efficient, and researchers are learning to run models on cheaper standard memory that costs a fraction of HBM. If demand shifts down-market, Micron's pricing power and 86% margins go with it. The point isn't that AI disappears — it's that Micron may not capture the benefit forever.

The matchup at a glance

Issue	Bull view	Bear view
Cycle	AI broke boom-bust	Industry nature is unchanged
Contracts	$100B and price floors locked	2027 supply offsets it
Margin	Durable pricing power	Priced for perfection, no slack
Valuation	Cheap at 9–14x forward	Low multiple on peak earnings
Demand	HBM shortage deepens	Efficiency shifts to standard memory

Where I land

Honestly, I think bull point #2 (the contract structure) and bear point #3 (whether HBM demand lasts) are the true center of gravity here. The $100B of contracts and price floors are a genuine safety net the old Micron never had. But those contracts don't guarantee today's ultra-high margins — a price floor stops the bottom from falling out, it doesn't defend an 85% margin.

So I grant the quality of this business while refusing to crown a winner today. In the next piece, I'll drop both scenarios into real numbers and calculate what Micron is actually worth.