Micron is about to be more profitable than any U.S. company except Nvidia and Google

Micron Technology has undergone a dramatic turnaround from losses three years ago to becoming one of the most profitable U.S. companies, trailing only Nvidia and Google. The key driver is surging AI demand for memory chips, with supply unable to keep pace and buyers willing to pay astronomical prices. The article points to a powerful improvement in Micron’s fundamentals and earnings power, though it does not provide specific quarterly figures.

Analysis

This is not just an MU earnings story; it is a pricing-power story created by a structural shortage in a single bottleneck input for AI compute. In the near term, memory is acting like a toll road on the AI capex boom: even if cloud spend slows, hyperscalers still need to secure supply, so the margin pool shifts from GPU vendors to the memory layer. The second-order implication is that suppliers with scarce, qualification-heavy components can sustain supernormal returns longer than the market expects because switching costs are high and inventory cycles are being replaced by allocation cycles.

The main loser is the customer, not the immediate competitors: hyperscalers and AI infrastructure builders will see model training economics worsen if memory remains tight, which can eventually pressure project ROI and capex cadence over the next 2-4 quarters. That said, the market may be underestimating how quickly Micron can transition from cyclical to quasi-oligopolistic earnings power if HBM and advanced DRAM capacity stays constrained through the next product cycle. The risk is that capacity additions by Micron and rivals are announced well before they show up in supply, so the stock can peak on forward supply expectations even while fundamentals stay strong.

Contrarian angle: the consensus may be treating this as a clean beneficiary trade, but the real question is duration. If AI memory ASPs are this elevated, the incentive for aggressive capex and technology substitution rises sharply, which can compress margins 6-12 months out. Also, if hyperscalers start optimizing around memory intensity rather than GPU counts, the relative benefit to NVDA could plateau while MU remains the cleaner earnings revision story in the medium term.

Near term, the setup is still bullish because earnings estimate revisions and gross margin expansion can outrun valuation compression for several quarters. But the higher the market prices this inflection, the more asymmetric the trade becomes into any sign of demand normalization, inventory buildup, or CAPEX deferral by cloud customers.

Key Decisions for Investors

• Long MU on pullbacks over the next 1-3 weeks; target a 15-25% upside move if margin guidance keeps stepping up, but trail a tight stop if memory pricing indicators roll over.
• Pair trade: long MU / short a diversified AI capex proxy over 1-2 quarters to isolate the memory bottleneck beneficiary from broader AI spend risk; best if you expect continued hyperscaler capex but slower multiple expansion elsewhere.
• Buy MU call spreads 3-6 months out to capture earnings revision momentum while limiting downside if the market starts discounting future supply additions.
• Take a relative-value long MU vs NVDA position for 1-2 quarters if you believe the next leg of AI economics shifts from compute scarcity to memory scarcity; NVDA remains strong, but MU has the more convex estimate-upgrade path.
• Set a profit-taking plan on MU if forward gross margin assumptions become consensus; the trade can reverse quickly once the street fully models the shortage and begins pricing in capacity normalization.