The Nasdaq Has Fallen in 48% of Septembers Since 1971. Here's What That Means for Nvidia and Micron.

The article highlights the “September Effect,” noting the Nasdaq’s long-run average return in September is about -1% (with downside moves typically sharper when it falls), driven by thinner liquidity and portfolio managers trimming winners/losses. It frames timing risk for AI bellwethers: Nvidia already reported in late August and could face post-earnings profit-taking at the start of September, while Micron reports fiscal Q4 earnings on Sept. 30 after potential seasonal selling. The key takeaway is not a thesis break for AI infrastructure/memory, but elevated risk of short-term drawdowns as positioning rotates.

Analysis

This is less a fundamental call on AI than a positioning air pocket in the most crowded parts of tech. When allocators trim risk into quarter-end and fiscal-year-end windows, the easiest source of cash is the highest-beta, most liquid winners, which means NVDA and MU can underperform even if the underlying demand story is intact. The second-order effect is factor damage: SMH and QQQ can lag broader growth as passive flows and de-risking reinforce each other, while lower-beta AI enablers like MSFT, GOOGL, and AMZN may hold up relatively better because they offer the same secular narrative with less single-name event risk.

The key catalyst path is not the calendar itself but the sequence of earnings and guidance. NVDA’s post-print digestion can set the tone for the next 2-3 weeks, but MU’s late-month report is the cleaner inflection point: if HBM pricing and supply constraints are still tight, the selloff should be faded because it confirms that AI capex is still supply-limited rather than demand-fading. Conversely, any sign of inventory normalization or margin pressure would justify a deeper de-rating in semis over the next 1-3 months. Watch whether QQQ breadth improves while semis lag; that would argue this is rotation, not thesis breakage.

The consensus is over-mapping a seasonal pattern onto a strong structural trend. September often hurts crowded winners, but crowding can also create an oversold snapback if macro data cooperates or if MU re-accelerates the AI memory narrative. The real falsifier is not a weak few sessions; it is a sustained break in AI capex commentary, guidance cuts, or a failure of semis to reclaim relative strength after MU. Absent that, this looks like a tactical drawdown, not the end of the AI trade.

Key Decisions for Investors

• Buy a 2-4 week SMH put spread financed partially by selling farther OTM puts; use it as a seasonal hedge into MU earnings and cover if SMH reclaims its pre-Sept relative high.
• Pair trade: long MSFT or GOOGL vs short NVDA for the next 2-3 weeks to express de-risking from crowded AI leadership into lower-beta AI exposure; stop if NVDA holds its post-earnings breakout on rising breadth.
• If MU sells off into the Sept. 30 print, look to buy the dip via a defined-risk call spread only on confirmation that memory pricing and HBM commentary remain tight; otherwise avoid bottom-fishing until the guide is public.
• Reduce gross exposure to QQQ/SMH into the first half of September, then re-add on any high-volume pullback that fails to break the month’s prior swing low; this is a positioning trade, not a conviction short.
• Use an alert on semis relative strength versus QQQ: if SOXX/SMH underperforms by more than ~3% over 5 trading days and breadth narrows, lean into the hedges; if relative strength stabilizes, treat the move as a tactical flush.