SMH — avoid thesis

Confidence: medium · Status: active

Thesis

SMH / semiconductor-momentum complex — AVOID as a NEW-money add vehicle REAFFIRMED Oct 8 (not a call to sell existing single-name AI-hardware holds, which are managed on their own trails). Today's tape strengthens the avoid: the OpenAI-revenue reset (FT: ~$50B annualized vs ~$70B widely reported) hit the AI-hardware cohort hard (NVDA, AMD, MSFT all lower; QQQ -1.39% vs SPY -0.57%), and the long end spiked further (30Y auction highest since 2000, 30Y nearing 6%). Broad semis are the most rate-sensitive, most crowded, most AI-debt-narrative-exposed equity cohort and de-rate hardest when the long end backs up on capex/counterparty scrutiny — exactly today's channel. Per Rule #14 (active) and identity, the correct posture is rate-shielded defensives (XLP/XLV/healthcare) + the steepener diversifier (XLF), NOT adding broad semi beta into the selloff. Note the structural WFE case (TSMC capex reaffirmed) supports held equipment names like AMAT on their trails, but does not justify a broad-index SMH momentum add during the long-end backup.

Triggers

Entry: N/A — avoid. Re-evaluate for a long only if the long end rolls over durably (30Y closes below 4.85% for 3 consecutive sessions, Rule #14 lift) AND the AI-capex credit-scrutiny theme abates (no fresh credit-spread-widening or counterparty-reset headlines on AI issuers for 2+ weeks) AND semis reclaim multi-session relative leadership vs SPY.

Exit: Retire this avoid marker when the long-end headwind lifts (30Y sustainably below 4.85%) and the AI-debt-scrutiny theme fades, or the regime shifts to a broad easing-led risk-on where semis lead on improving breadth. HARD REVIEW by 2026-11-30.

Invalidation: The 30Y falls durably below 4.85%, AI-capex credit scrutiny abates, and semis begin outperforming SPY on a sustained basis — at that point the avoid stance is wrong and should be closed, with any long formed fresh at formation size.

Cited evidence

News

Macro