How Micron Could 3x From Here If AI Memory Demand Keeps Exploding

The post How Micron Could 3x From Here If AI Memory Demand Keeps Exploding appeared first on 24/7 Wall St..

Three times the current price would put Micron Technology (NASDAQ:MU) somewhere around $3,000 a share. That is a number that sits above every published Wall Street target on the stock. The only way there runs through a memory supercycle that lasts longer and runs hotter than even bulls currently model. Let’s take a look at how Micron could get there, even though Goldman Sachs has a sharp counter to the whole exercise.

The June quarter that re-rated DRAM and HBM

Micron’s fiscal third quarter, reported June 24, forced analysts to redraw their spreadsheets. Revenue came in at $41.456 billion, up 345.72% year over year from $9.30 billion. Non-GAAP EPS landed at $25.11 against a $20.2843 consensus. GAAP gross margin printed 84.6%, against 37.7% a year earlier. Operating cash flow of $25.388 billion in a single quarter argues for structural change beyond a normal cyclical bounce.

CEO Sanjay Mehrotra was direct about why. “The memory industry has been structurally transformed by the proliferation of AI.” Management has signed 16 strategic customer agreements with take-or-pay terms that represent roughly $100 billion in remaining performance obligations and projected $22 billion in customer cash deposits and related commitments.

The strategic supply agreement with Anthropic disclosed June 22, 2026 is the headline example. Mehrotra also said HBM can fill only 50% to two-thirds of demand in 2026.

The bull thesis writes itself from there. Micron is one of only three HBM suppliers, alongside SK Hynix and Samsung. It crossed $1 trillion in market cap in May. Year to date the stock is up 296.92%. Q4 guidance of $50 billion in revenue and $31.00 in non-GAAP EPS implies the trajectory is accelerating.

What a 3x from here actually requires

At roughly $1,046.96 on June 29, a 3x outcome puts shares near $3,000. Every sell-side target sits below that. UBS is the Wall Street high near $1,625. Bank of America, Needham, Stifel and TD Cowen cluster between $1,300 and $1,600. Post-earnings, Susquehanna and DA Davidson have pushed to $2,000, which is the most aggressive published number on the desk and still well short of the headline scenario.

To get to $3,000, you need a stack of three things to break right. First, EPS power has to climb past the $31.00 ± $1.00 Q4 guide into a sustained run rate few analysts will underwrite today. Second, the multiple has to hold or expand, even though the trailing P/E already sits around 26x and the forward P/E near 7x reflects estimates that have not yet caught up to guidance.

Third, the take-or-pay contracts have to do what Mehrotra claims, which is convert memory into something closer to a utility deserving a premium for predictable cash flows. Each of those is defensible. Obviously, none is guaranteed.

The bear case Goldman is making

Goldman Sachs sits at roughly $900. Their argument is that today’s near 85% gross margins represent a cycle peak that will normalize lower. Memory has always been a synchronized capacity business, and SK Hynix, Samsung and Micron are all spending.

Micron itself guided full-year FY2026 capex to approximately $27 billion, with ID1 on track for first wafer output in mid-calendar 2027 and ID2 in late-calendar 2028. When that bit supply lands, even strong AI demand will compress pricing power. CFO Mark Murphy effectively conceded the point on the call, saying “we are at margin levels where incremental price yields less gross margin expansion.”

There is also concentration risk. Lead-customer dependence on HBM4, hyperscaler research into memory compression techniques that could cut usage by up to 40x, and the historical track record of memory cycles ending badly all sit on this side of the ledger.

The verdict

The fair read is that Micron is a different company than it was 24 months ago, and the multi-year contracts genuinely change the cyclicality argument. A double over a few years is defensible on the numbers in hand.

A triple requires the supercycle to extend deep into 2028 and beyond, with margins holding far above prior peaks. Bulls have current data. Bears have cycle history. Both are right until one of them is not.

The post How Micron Could 3x From Here If AI Memory Demand Keeps Exploding appeared first on 24/7 Wall St..