Burry says the AI bubble may pop early and Micron is this week's big test — buy downside options on MU into earnings

Verdict: The China thesis is coherent, but the fade trigger has not arrived — wait

This is a bearish fade-the-rally idea on NVDA and MU built around China's chip challenge — Huawei share gains (per Reuters, October 13, 2026), China's export surge (per Bloomberg, October 14, 2026), and the CXMT memory rout (per CNBC, October 13, 2026). The strongest point for it is real: provisional ownership filings for the period ended September 30, 2026 show net open-market insider selling of roughly -$569.5M at NVDA and -$231.1M at MU, and NVDA's net margin fell 9.4 points to 62.0% while free cash flow dropped 56.0% to $21.4B in the quarter ended July 26, 2026. The strongest point against it is that the reported numbers still look like a boom — MU's quarter ended September 3, 2026 delivered revenue up 30.8% to $54.2B with an 86.8% gross margin — and the strategy itself has fired exactly once in nine months, a single +36.1% trade with a 14.0% drawdown filled on daily bars, so the execution evidence is one favorable episode, not validation. Nothing triggers today: NVDA's RSI (14) reads 47.1 against the 65 threshold, and MU sits at RSI 24.9, $38.12 below its 50-day EMA. Verdict: wait — no position until NVDA's RSI crosses above 65 (or MU closes above its 50-day EMA and upper band with RSI above 65), and reassess the entire thesis on the next NVDA earnings print and the confirming insider-filing cycle. A China revenue re-acceleration at NVDA, or a confirmed reversal of that provisional insider selling, would flip the verdict.

Trade now

Nothing triggers today, and that is the actionable message. The strategy is a fade-the-rally entry: it goes long only when a name is stretched — price crossing above its 50-day EMA, closing above the upper 20-day Bollinger band, the 14-day RSI above 65, with the MACD histogram below its signal line. NVDA is closest: at $228.52 it already sits above its 50-day EMA of $222.78 (a $5.74 gap) and above its upper Bollinger band of $228.34, and the MACD condition is met — but the 14-day RSI reads 47.1, needing a move of roughly 18 points to clear 65. On MU the setup is further away on every leg: the last close of $953.19 sits $38.12 below its 50-day EMA of $991.31 and $87.52 below its upper Bollinger band of $1040.71, with the 14-day RSI at 24.9 — more than 40 points below the trigger. Waiting, concretely, means: no position until NVDA's RSI crosses above 65, or MU closes above both its 50-day EMA and upper band with RSI above 65. Risk is defined by the rules, not by feel. Once in, a position exits on signal reversal (price below the 50-day EMA with RSI below 50 after 45 bars), at the first support level, at a hard stop of about -2.5%, or at a take-profit of about +5.1% — roughly a 2:1 reward-to-risk envelope per trade, sized via fixed risk at about 2.5% per position with a 25% maximum position weight. The completed backtest on the NVDA pair over the last nine months produced one trade with a +36.1% return and a 14.0% maximum drawdown, a profile consistent with patient, risk-bounded fading. Treat those fills as approximate — exits were filled on daily bars, not intraday. One setup note: the parameter search could not be completed because market-data coverage was incomplete, so no robust alternative parameter setup was established — trade the rules as published, unmodified. If NVDA rallies into the trigger zone on a China-competition headline, the entry conditions can all align within days; if MU keeps bleeding (RSI in the mid-20s), the long-side trigger stays out of reach and patience remains the position.

China's Chip Push Is Hitting the Numbers, Not Just the Headlines

The bearish thesis rests on a supply-side shift that is now visible in the tape, not just in commentary. Per the Reuters piece from October 13, 2026, Nvidia's China AI-chip share is falling as Huawei closes the gap, and Bloomberg's October 14 coverage of China's 'extraordinary' trade surplus frames an export machine that now includes AI chips shipping at nearly double prior rates. CNBC reports that CXMT's memory-chip progress already triggered a global stock rout — meaning the market has demonstrably repriced Chinese competition once, and can do it again. The fundamentals add a warning sign at Nvidia specifically. In its quarter ended July 26, 2026, net margin fell 9.4 percentage points sequentially to 62.0% even as revenue rose 17.9% to $96.2B — a sign that pricing power, the core of the AI-chip moat, is beginning to erode. Free cash flow dropped 56% quarter over quarter to $21.4B, and debt-to-equity more than tripled from 0.04 to 0.15. If Chinese competition compresses Nvidia's China revenue — the line item the idea flags as most exposed — those margins have the most room to fall from here. The idea's execution evidence is a completed 9-month backtest: one trade, a win, a 36.1% return against a worst drawdown of 14.0% on a daily setup in NVDA and MU. The trade sits alongside a clear ownership signal: in the provisional ownership report for the period ended September 30, 2026, insider direction was net open-market…

Scores

• Conviction score breakdown: 46
• Thesis support: 60
• Trade readiness: 35
• Risk quality: 55
• Backtest evidence: 30
• Fundamentals trend: 50

Watch items

• NVDA — RSI (14)
• NVDA — Upper Bollinger band (20-day)
• MU — RSI (14)
• MU — Price vs 50-day EMA
• NVDA — Insider net open-market selling (provisional, period ended 2026-09-30)
• NVDA — Free cash flow, quarter over quarter change
• MU — Deepest listed support level