The Club's top 10 things to watch in the stock market Thursday

Micron delivered a strong quarter, with EPS of $25.11 versus $20.86 expected and August-quarter EPS guidance of $30-$32 above the $25.72 consensus, while warning that memory-chip supply remains tight through beyond 2027. Qualcomm also rallied after lifting its fiscal 2029 non-handset revenue target to $40 billion and targeting more than $15 billion from AI data center infrastructure, including a Meta CPU deal. Banks passed the Fed stress tests and boosted capital returns, while Darden and McCormick posted mixed-to-solid results and Intel, Alphabet, and Affirm were pressured by analyst actions or talent concerns.

Analysis

The biggest second-order takeaway is that AI demand is now creating a visible pricing and allocation regime across the semiconductor stack, and that favors the names with either proprietary compute, scarce supply, or binding customer relationships. The market is likely underestimating how much tighter memory availability can support not just the direct beneficiary, but also adjacent capex spend in networking, HBM-adjacent equipment, and foundry capacity over the next 6-12 months. That said, the stronger the demand signal, the more it raises the odds of an eventual digestion phase if customers pre-buy aggressively into 2026.

Qualcomm’s message is more important than the stock reaction: it is attempting to re-rate from handset beta to diversified AI infrastructure exposure. If that transition gains credibility, the multiple gap versus NVDA/AVGO narrows, but execution risk is high because investors will demand proof that data center revenue is durable rather than one-off design wins. META’s involvement helps validate the thesis, yet it also highlights concentration risk if hyperscalers keep dual-sourcing and pressuring economics.

The bank stress-test outcome is a quiet positive for capital return, but the real implication is that buybacks become a more reliable support layer in a market that already has crowded growth ownership. JPM’s incremental payout power is the cleanest signal; GS benefits more from deal activity than from balance-sheet expansion. In consumer, the soft guidance from DRI suggests high-income dining is stabilizing, not accelerating, which is a warning sign for discretionary names that rely on premium traffic continuing to outrun inflation.

Alphabet’s research-exit headlines matter less for near-term earnings than for the marginal increase in competitive fragility around frontier AI talent. The market may be underpricing the risk that talent drain slows internal model cadence just as product competition intensifies, which could keep the stock range-bound for months even if ad fundamentals remain intact. AFRM looks tactically stretched; after an 80%+ rebound, the easy part of the rerating is likely over, and any rate or consumer-credit wobble could trigger a fast multiple reset.

Key Decisions for Investors

• Stay long NVDA and AVGO; use any QCOM strength to add only on pullbacks. Framing: QCOM is a re-rating story that needs quarters of proof, while NVDA/AVGO still own the highest-conviction AI revenue capture over the next 2-4 quarters.
• Buy QCOM only as a relative-value long against INTC or a basket of legacy semi names, not outright. Risk/reward improves if the market starts paying for AI infrastructure optionality, but the stock likely needs 1-2 more catalysts to sustain a higher multiple.
• Add JPM on post-stress-test weakness and pair it against lower-quality financials. The buyback/dividend support should cap downside over the next 1-3 months, and capital return visibility is now a stronger floor than earnings upgrades.
• Fade AFRM into strength with a tactical short or put spread over 1-2 months. After the move from March lows, upside now requires sustained credit quality and consumer resilience, while the downside can reprice quickly if rates or delinquencies move against it.
• Use any further GOOGL weakness to wait, not buy aggressively, until talent retention stabilizes. Near-term sentiment is vulnerable to AI-credibility headlines, so the better trade is patience rather than catching the falling knife.