Michael Burry Just Revealed His Next Big Short, and It's a Bet Against Nvidia, Micron, and AMD

Michael Burry disclosed he is short the iShares Semiconductor ETF (SOXX) as of June 30, betting that AI infrastructure spending is unsustainable and could trigger a correction. The ETF trades at a high P/E of 74.3 (more than 2x the Nasdaq-100), while Burry argues customer-side AI returns are weaker and a UBS survey suggests ~60% of businesses are curbing AI spending. With SOXX up sharply during the AI boom (annualized >54% over the past three years), the trade is positioned for downside if spending slows further.

Analysis

The disclosure matters more as a sentiment and positioning catalyst than as a fresh fundamental datapoint. SOXX is a high-beta proxy with crowded ownership and a concentrated mix, so even a modest re-rating can force systematic de-risking faster than earnings estimates change. Near term, the risk is multiple compression first, fundamentals second; the market can punish the whole basket on any sign that AI budgets are becoming more selective.

The most vulnerable legs are the names with the highest operating leverage to incremental AI spend: memory and GPU exposure should react first if hyperscaler capex moderates. Over 1-3 months, that means MU, NVDA, and AMD can underperform if customers stretch deployment schedules or push out incremental orders; over 6-18 months, the bigger issue is whether AI spend shifts from brute-force scaling to efficiency-driven architectures. That shift would be less negative for AVGO and TSM than for merchant GPU suppliers, because custom silicon and supply-chain bottlenecks can preserve demand even if total spend slows.

The contrarian miss is that a slowdown in growth rate is not the same as a collapse in absolute demand. If capex remains elevated and only the mix changes, the short can work on valuation but fail on earnings, which is a dangerous setup for a crowded bearish trade. The key falsifiers are unchanged/upward hyperscaler capex commentary, resilient HBM pricing, and continued backlog strength into the next earnings cycle; if those hold, this becomes a squeeze-risk trade rather than a structural short.

Key Decisions for Investors

• Short SOXX or buy SOXX put spreads into the next round of hyperscaler earnings/capex commentary; target a 1-3 month window where valuation compression can outrun earnings revisions. Risk: repeated capex guidance upgrades would likely squeeze the short quickly.
• Relative-value: long AVGO / short MU for 1-3 months. Thesis: if AI spend becomes more ROI-disciplined, custom silicon and networking should hold up better than HBM memory. Falsifier: another leg of HBM price tightening or an upside MU guide.
• If expressing the bear case more directly, use a short NVDA vs long TSM or AVGO hedge rather than a naked short in single names. This reduces factor risk and isolates the thesis that merchant GPU multiples are most exposed to a capex pause.
• Avoid chasing downside after a single red session; wait for confirmation from earnings/guidance revisions. If SOXX fails to break back above recent momentum support after the next major capex print, that is the cleaner entry for bearish exposure.
• Set a stop-loss on the thesis, not just the trade: if hyperscaler capex remains flat-to-up for two consecutive quarters, cover bearish semiconductor exposure and rotate to relative longs in AVGO/TSM.