Wall Street Says Micron Is the Next Nvidia. The Business Model Says Be Careful.

Micron just had the kind of week that makes traders forget their own history. The Boise, Idaho memory maker briefly passed Meta and Tesla in market value on Thursday, its stock is up more than 236% in a single month, and Wall Street has decided it has found the next Nvidia. The earnings backed up the hype: fiscal Q3 revenue of $41.46 billion, up 346% from a year ago, with gross margins near 85%.

The numbers are real. The “next Nvidia” framing is where founders and investors should slow down. Micron and Nvidia are riding the same AI wave, but they are not the same kind of boat, and the difference is the entire ballgame.

High-Bandwidth Memory (HBM) is a type of DRAM stacked vertically and wired directly next to an AI processor to feed it data fast enough to keep up. A single AI server needs orders of magnitude more memory than a laptop, which is why the AI buildout has turned a sleepy commodity into the tightest supply chain in tech.

What actually happened

Micron’s fiscal third quarter did not just beat estimates. It detonated them. Here is the scoreboard.

Then management guided fiscal Q4 revenue to roughly $50 billion, several billion above what analysts expected, and projected gross margin near 86%. For a company that spent years stuck below $100 a share before mid-2025 and closed Friday at $1,132, this is a genuine regime change.

The driver is RAMageddon: the AI data center boom has created a shortage of both DRAM and NAND memory, with HBM the scarcest and most profitable slice. Nvidia and the hyperscalers (Microsoft, Amazon, Google, Meta, Oracle) are hoovering up supply, which forces everyone else, from Dell and HP to phone makers, to hoard whatever is left. That shortage is forecast to persist into 2027, and Micron’s CEO told analysts the tightness is “locked in” beyond calendar 2027.

The bull case is better than the headline suggests

The lazy version of the bull case is “AI demand is huge.” That is true but useless, because it was also true in every prior memory cycle right before the bust.

The real bull case is structural, and it is the one thing that could make this time different: Micron signed 16 strategic customer agreements carrying roughly $100 billion in minimum contracted revenue and about $22 billion in upfront customer cash, including deals with Nvidia and Anthropic. These are take-or-pay style commitments. If they hold, they convert a spot-priced commodity into something that looks more like a contracted backlog, which is exactly the kind of revenue visibility that justifies a higher multiple. The CFO’s framing was blunt: committed volume lets Micron invest with confidence.

That is the actual thesis. Not “AI is big.” It is “Micron may have engineered its way out of the boom-bust trap that has defined memory for forty years.”

Why “the next Nvidia” is the wrong mental model

Here is where the skeptic earns their keep. Nvidia and Micron win on completely different terrain.

Nvidia’s pricing power comes from a software and architecture moat that competitors have spent a decade failing to cross. Micron’s current pricing power comes from a supply shortage. Those are not the same asset. A shortage is a clock, not a castle. The moment supply catches demand, the commodity reprices, and Micron does not own a CUDA to fall back on.

It is worth saying plainly: in the highest-value part of this boom, HBM for Nvidia systems, Micron is the number three vendor. SK Hynix holds the leading position. Calling the third-place commodity supplier “the next Nvidia” is a stretch the earnings do not actually support.

The number to watch is the margin, not the revenue

Everyone is staring at the revenue line. The tell is the 84.9% gross margin. Memory businesses historically run 20% to 40% gross margins. An 85% margin in a commodity is not a moat, it is a scarcity rent, and rents get competed away. Micron, SK Hynix, and Samsung are all racing to add capacity. Micron broke ground in January on a roughly $100 billion fab complex in Clay, New York, but those facilities do not deliver meaningful output until around fiscal 2028.

That is the whole story in one sentence: the bull case is that demand outruns new supply through 2027, and the bear case is that three giant manufacturers adding capacity at the same time eventually do what they have always done, which is build a glut. Micron even flagged a “meaningful moderation in the rate of price increases” in its Q4 outlook. That phrase is the first small crack worth tracking.

The Business Model Analyst Take

Micron is a phenomenal business having a phenomenal moment, and the take-or-pay contracts are a legitimately clever attempt to break a forty-year cyclical curse. Give the management team real credit for that.

But “the next Nvidia” is a story Wall Street tells because it is eager for one, not because the business models rhyme. Nvidia sells a moat. Micron sells a commodity into a shortage. Those can both make you rich, but only one of them protects you when the cycle turns, and memory cycles always turn.

For founders, the lesson is the cleaner one: when your margins triple because the market is starved, do not confuse the scarcity for a moat. Micron is using a windfall to buy structural protection through long-term contracts. That is the right move, and it is the part of this story actually worth copying. The stock chart is not.

Reporting based on Micron’s fiscal Q3 2026 results, TechCrunch, CNBC, and Micron’s earnings call.

Frequently Asked Questions