The Inflation Number Warsh Is Actually Watching

Micron delivered blowout results, with quarterly revenue of $41.5B versus $9.3B a year ago and adjusted EPS of $25.11, well above the $20.28 consensus; gross margin surged to 84.9%. Management guided next-quarter revenue to about $50B, far above the $43.6B estimate, and said AI-driven memory demand is likely to keep supply tight beyond 2027. The article also notes May PCE inflation at 4.1% headline and 3.4% core, while the Dallas Fed trimmed mean PCE held near 2.4%, reinforcing a more dovish underlying inflation signal.

Analysis

The clearest read-through is not “AI is strong,” but that the memory cycle is being repriced from cyclical to quasi-contractual. If a meaningful share of Micron’s future revenue is locked under multi-year take-or-pay style arrangements, the market should start valuing the memory stack less like commoditized silicon and more like constrained infrastructure capacity. That is constructive for the whole AI capex complex: it lowers the probability of a near-term inventory air pocket and supports continued ordering from hyperscalers, which is a tailwind for the large cloud platforms and their equipment ecosystems.

Second-order effects matter more than the headline beat. Persistent memory tightness tends to push customers toward design wins that maximize performance per watt and per byte, which should favor the most entrenched cloud vendors and penalize smaller AI builders with weaker negotiating leverage. It also raises the odds that “AI revenue” quality bifurcates: the firms with real end-demand and scale economics can absorb higher input costs, while the marginal players may need more vendor financing, credits, or circular demand to sustain growth.

On inflation, the key market implication is that a sticky headline/core print does not automatically force a hawkish regime shift if the trimmed-mean series stays near 2.4%. That keeps the Fed path closer to “higher for longer, but not reacceleration,” which is supportive for long-duration growth as long as rates do not reprice materially higher. The risk is a two-month sequence of upside in trimmed mean to the high-2s, which would undermine the disinflation narrative and hit crowded AI multiple names hardest.

The contrarian takeaway is that today’s AI trade is less about multiple expansion than about whether earnings can keep compounding without being contaminated by circular capex loops. If the market begins discounting cloud backlog quality, the highest-beta beneficiaries could de-rate faster than the ecosystem leaders. In that case, the trade is not to fade AI outright, but to own the franchises with genuine operating leverage and short the names most dependent on financial engineering or third-party demand substitution.

Key Decisions for Investors

• Long MU vs. short a basket of weaker AI memory consumers for 1-3 months: use MU as the quality beneficiary of structural supply scarcity; hedge with short exposure to high-beta hardware names that lack pricing power. Risk/reward: asymmetric if memory contracts stay tight into next quarter.
• Add GOOGL and MSFT on pullbacks over the next 2-4 weeks, but size below normal until trimmed-mean PCE confirms sub-2.5% stability. These are the best-positioned AI cash generators if inflation remains contained and capital intensity is still rewarded.
• Initiate a pair trade long MSFT / short ORCL for 2-6 months. MSFT has better cloud scale and optionality if AI demand is real; ORCL is more exposed to backlog-quality skepticism if the market starts questioning how much growth is circular.
• Buy 3-6 month call spreads on NQ or QQQ after any rate-driven selloff: the setup favors long-duration growth if trimmed mean stays anchored and the Fed does not turn meaningfully more hawkish. Use spreads to reduce theta given elevated AI crowding.
• Avoid chasing NDAQ upside here; prefer using any Nasdaq strength to trim or hedge. The index can stay bid, but if the market re-rates AI earnings quality, the first air pockets should hit benchmark-heavy exposure before the fundamentals story fully breaks.