"Big Short" Investor Michael Burry Is Now Betting Against Micron, Nvidia, and Tesla. Should You Be Worried?

Michael Burry disclosed new short positions against Nvidia, Tesla, Applied Materials, Caterpillar, SOXX and added Micron shorts near ~$1,052/share, arguing AI-linked semis are priced at historical extremes. In the same news flow, Micron reported fiscal Q3 revenue of $41.5B (up >4x YoY) and guided to ~$50B in fiscal Q4 revenue, while Tesla delivered 480,126 vehicles (+~25% YoY) but trades at >350x trailing earnings with a 4.2% Q1 operating margin. The article’s core takeaway is a cautious valuation/earnings-cycle stress test rather than evidence of immediate business failure.

Analysis

This is less a fresh fundamental shock than a positioning event in a crowded factor complex. In the next few days, the main transmission channel is dealer/hedge-fund de-grossing: when an iconic short frames the AI trade as a consensus crowding problem, the weakest holders tend to sell first, and the highest-beta semis usually trade on flow before earnings. The key is dispersion — the basket is not one trade, and the market often overreacts by pricing every AI-linked name as if capex demand were identical.

The most vulnerable names are the ones whose valuation depends on second-order monetization or peak-cycle economics, not current cash generation. TSLA is the cleanest example: if the market has to wait another few quarters for autonomy/software to matter, the multiple can compress faster than unit growth helps. MU and AMAT are more cyclical than they look on a one-quarter snapshot; if memory pricing or fab spending normalizes, the earnings power gets marked down very quickly because the market pays for cycle peak, not the reported run-rate.

The contrarian miss is that NVDA is not the same animal as the rest of the basket. It is the least dependent on perfect execution from the downstream ecosystem, so an outright short there is usually a timing trade, not a valuation trade. Over 1-3 months, the real falsifiers are hyperscaler capex guides, memory ASP/inventory commentary, and any Tesla update that advances monetization timing; over 6-18 months, the risk is not collapse but multiple compression as AI spend matures and investors demand free cash flow durability instead of narrative optionality.

Key Decisions for Investors

• Use any AI-led selloff to go long NVDA vs short SOXX on a 1-3 month horizon: NVDA has the strongest earnings quality in the basket, while SOXX carries more beta to a capex digestion phase. Risk/reward favors the pair as long as data center revenue keeps outgrowing the broader semi cycle; exit if hyperscaler capex guides flatten or NVDA margin expansion stalls.
• Avoid outright shorting NVDA; if expressing the bearish view, use a smaller hedge ratio or a call spread overlay instead of a directional short. The trade only works if next quarter’s guide shows clear deceleration, otherwise the stock can stay expensive for longer than most short books can remain solvent.
• For TSLA, consider a bearish structure only on rallies — preferably a put spread into event/earnings volatility rather than spot shorting. The position is highest convexity if robotaxi/software monetization slips by 1-2 quarters; invalidate if management pulls forward a credible revenue timeline or operating margin re-accelerates.
• On MU, wait for confirmation from memory pricing and inventory data before pressing the short; if those metrics stay tight, the stock can remain mechanically supported despite cycle risk. Best expression is a short-dated put spread around the next earnings window, with thesis failure if ASPs and contract commentary stay resilient.
• Treat AMAT and CAT as second-order beneficiaries/losers of AI capex, not pure AI beta: if hyperscaler spending guides down, AMAT is the cleaner short than CAT because equipment orders reprice faster than general industrial demand. Use only after confirmation from capex commentary; otherwise the lag can be painful.