Is It Too Late to Buy Micron Stock After Its 981% Surge?

Micron (MU) is up 981% since the start of 2025 on an AI-fueled memory boom, but investors are questioning sustainability as margin growth slows: non-GAAP operating margin rose to 81.2% in fiscal Q3 from 26.8% a year earlier, yet Micron expects non-GAAP gross margin to increase by just 1% sequentially to 86% in fiscal Q4. The article argues secular DRAM demand remains stronger than supply (Citrini Research estimates a 28.7 exabyte demand shortfall in 2030), with long-term agreements expected to generate at least $100B in revenue and analysts’ 12-month median price target of $1,600 implying ~75% upside.

Analysis

The setup is better viewed as a transition from pure multiple expansion to cash-flow durability. When a memory name gets this far ahead of itself, the first thing the market sells is not earnings power but the rate of improvement; that makes MU vulnerable to short-term de-rating even if absolute profits keep rising. The likely near-term losers are the broad semi basket and any AI hardware name that depends on tight memory allocation, because investors will start asking whether constrained supply is an input cost issue for the next leg of AI deployment.

The more interesting second-order effect is that capacity additions do not immediately solve pricing, but they do change who captures the next dollar of demand. Over the next 1-3 months, the key catalyst is not headline demand; it is whether contract pricing and mix continue to improve faster than consensus expects. If they do, MU stays a quality cyclically-extended compounder; if not, the stock can still work on revenue growth, but the multiple is exposed to compression as the market stops paying for peak-margin momentum.

Contrarian take: the crowd is likely overfocusing on the slowing margin slope and underappreciating the visibility created by long-term supply agreements. That said, the market may also be underestimating how quickly competitors will chase returns once margins stay elevated, which caps the duration of supernormal economics. The thesis is falsified if DRAM/HBM pricing rolls over for multiple months or if management signals that 2027-2028 gross margin expansion is flattening sooner than expected; in that case, the right move is to fade strength rather than buy the dip.

Key Decisions for Investors

• Buy MU on a post-rally pullback or into earnings weakness; 3-6 month horizon. Risk/reward favors owning the name if margin guidance remains constructive, but trim if the stock re-rates purely on earnings momentum without new contract evidence.
• Pair trade: long MU / short SMH for a 1-3 month relative-value expression. The idea is to isolate memory scarcity and contract visibility versus the broader semi basket, where multiple risk is higher if AI enthusiasm cools.
• Do not chase MU after an outsized up-day unless spot DRAM and HBM pricing data confirm continued tightness. If pricing flattens for 2 consecutive monthly checks, take profits or reduce exposure.
• Watch NVDA and AI server supply chains as a second-order alert, not a primary hedge: if memory remains the bottleneck, AI deployment timing can slip, which can pressure the whole AI hardware complex even if MU itself holds up.
• If you need optionality, use a call spread rather than outright calls after a large move; the asymmetric risk is that peak-cycle multiple compression can offset decent earnings growth over the next 1-2 quarters.