AI stocks tumble as the defensives rise. What it says about this uneasy market

Stocks fell Thursday as renewed selling in AI-related shares followed Jensen Huang's comments about neoclouds' scarcity of capital, Meta's AI funding needs, and Lam Research's weaker current-quarter guidance; GE Vernova dropped almost 10% and Meta fell nearly 6%. Investors rotated toward healthcare and consumer staples while Treasury yields moved lower despite rising oil prices. The session reversed some of the prior day's gains in AI-linked stocks, while Johnson & Johnson gained over 2%.

Analysis

The selloff is exposing a distinction the AI narrative has blurred: end-user demand can remain strong while the capital providers and project economics become the bottleneck. That is most adverse for capital-intensive neocloud exposure such as CoreWeave and for suppliers whose near-term orders depend on customers converting ambitious build plans into funded deployments. Lam Research’s guidance adds a separate execution check: AI demand does not immunize the semiconductor supply chain from weaker near-term spending. GE Vernova’s sharp reversal suggests investors are treating data-center power demand as a long-duration theme, not a near-term earnings floor.

For Meta, the proposed non-voting share class raises a financing and governance overhang, not proof that the AI opportunity is impaired. Watch the eventual issuance terms and capex trajectory: equity funding could dilute existing holders, while debt or internal funding could pressure returns or other spending. The market’s reaction will depend on whether incremental AI investment earns returns commensurate with its cost.

Near term, risk-off flows can sustain relative strength in staples and healthcare, but one session is not evidence of a durable rotation. Over 1–3 months, monitor Meta’s capex/funding detail, Lam Research’s order outlook, and evidence that neocloud customers can finance capacity. Over 6–18 months, power and compute constraints may still benefit infrastructure suppliers if projects are funded and delivered. The contrarian risk is extrapolating a funding bottleneck into an AI-demand collapse; conversely, demand commentary alone is not a substitute for funded orders and returns.

Key Decisions for Investors

• Consider a modest, time-limited relative-value hedge: long XLP versus short SMH, preferably after the initial risk-off move stabilizes. This expresses defensive rotation without assuming AI demand disappears. Reassess if semiconductor guidance/order trends stabilize and the spread reverses; keep sizing conservative because a rebound in yields or AI risk appetite could unwind the trade quickly.
• Do not chase GEV weakness solely on the day’s move, or buy CRWV on demand rhetoric alone. For CRWV, wait for evidence on customer funding, financing terms, and utilization; for GEV, look for funded data-center power orders and delivery visibility. A sustained deterioration in either would invalidate the structural-demand case.
• Keep META on watch rather than buying the dip automatically. Verify the new share class’s issuance terms, dilution potential, and management’s capex/return framework. Improving funding clarity alongside disciplined investment would reduce the overhang; rising spending without credible return evidence would argue against adding exposure.
• Treat KMB and JNJ as relative defensives, not risk-free substitutes. KMB’s pending Kenvue transaction adds execution and financing variables; monitor deal terms and integration guidance. A renewed rise in yields or weak staples guidance would challenge the rotation thesis.