AI stocks plunge. Inflation heats up. Where we're still finding opportunities

The Nasdaq fell 4.5% last week, its worst weekly decline since June, while the S&P 500 lost 2%; the Dow gained 0.7% as investors rotated into less AI-dependent sectors. September CPI rose 3.7% year over year versus 3.6% expected, and PPI increased 1.3% for the month and 6.3% annually, renewing concerns about persistently high rates and AI funding costs. Semiconductor stocks and cybersecurity holdings slumped, while Meta and Microsoft held up comparatively well; bank earnings were mostly solid, though results did not generally lift share prices.

Analysis

The market’s key distinction is shifting from “AI exposure” to “who can finance it and monetize it.” Persistent inflation would pressure long-duration valuations broadly, but the sharper second-order risk is a higher cost of capital meeting infrastructure plans that require sustained external funding. Credit spreads and refinancing access are therefore more informative than a one-week equity drawdown for CoreWeave (CRWV) and the AI supply chain. If financing tightens, spending delays could spill from neoclouds into semiconductors and storage; the memory-competition concern adds a separate risk that supply growth compresses pricing even if AI demand remains intact. Verify CXMT’s actual production scale and customer qualification before treating its technology announcement as a near-term earnings threat to Micron (MU), Sandisk (SNDK), Western Digital (WDC), or Seagate (STX).

The contrarian opportunity is that indiscriminate AI de-risking may conflate leveraged infrastructure with platforms that can fund investment from established businesses. Microsoft (MSFT) and Meta (META) may be more resilient, but product claims and model rankings are not proof of incremental monetization; watch cloud capacity utilization, AI-driven revenue, and spending growth. Cybersecurity weakness also appears partly narrative-driven: an early-stage acquisition discussion is not evidence that Palo Alto Networks (PANW) or CrowdStrike (CRWD) face immediate demand erosion. Still, competitive substitution could pressure future pricing if AI providers bundle security capabilities.

Over 1–3 months, inflation and credit conditions are the main catalysts; over 6–18 months, funding discipline and memory supply determine whether AI capex converts into durable returns. Bank results do not remove macro risk: higher rates can support some income streams while increasing credit and funding pressure. BNY’s fee-based model is a relative diversifier, though market declines can reduce assets under custody.

Key Decisions for Investors

• Favor a measured quality-versus-financing-risk pair: long MSFT against a smaller short or underweight in CRWV, rather than shorting the AI theme wholesale. Enter only if CRWV credit protection or bond yields remain stressed; reduce the hedge if spreads stabilize and financing access improves. Falsifier: sustained improvement in CRWV credit pricing alongside evidence that spending is supported by operating cash flow.
• Keep MU, SNDK, WDC, and STX exposure below benchmark pending evidence on CXMT’s commercial volumes and industry supply discipline. Reassess on actual customer qualification, capacity additions, and memory pricing—not the announcement alone. A halt in supply expansion or resilient pricing would weaken the bearish thesis.
• Do not chase the cyber sell-off solely on the reported OpenAI discussions. Monitor deal confirmation, product scope, and PANW/CRWD bookings or guidance for evidence of substitution; absent that evidence, treat the move as a watchlist opportunity rather than a confirmed earnings downgrade.
• Use BNY as a potential relative diversifier, not a macro hedge: track fee revenue and custody assets alongside market levels. A material deterioration in fee trends or a renewed broad risk-off move would invalidate the relative-resilience case.