Micron Just Priced AI Memory to 2031—And Burry's $205 'Cycle Peak' Short Is Now Underwater 424%

Every bear thesis on memory chips since 2024 has rested on the same assumption: AI demand fills the order book for two to three years, then a capacity wave meets a maturing build-out and pricing collapses. Micron Technology's September 30 fiscal Q4 report stretched that demand window into 2031, with multi-year take-or-pay agreements, $32B in upfront customer cash deposits and roughly $150B of remaining performance obligations priced into the backlog. The cycle has not peaked on a 24-month horizon — it has been contractually pushed into 2031.

Michael Burry opened a direct short against Micron Technology at $205.10 on July 2, 2026 — and added to it in late September at roughly $1,051 per share as the stock kept climbing. With Micron now at $1,074.89, his original position is underwater 424% relative to cost, and his average entry is now nearly identical to the current price. The thesis — a 'managed surplus' built on conventional-DRAM normalization, Chinese capacity additions and a 42-year memory drawdown history — is being invalidated in real time by the September 30 print.

The structural rebuttal matters more than the P&L. Burry's 42-year history relies on cyclicality driven by capacity build-then-glut — but Micron Technology's 35%-of-reverage-through-2030 SCA coverage plus $32B in non-refundable customer deposits changes the cash-flow profile of the company itself. Negative free cash flow in 48% of historical quarters required external capital; the FY2026 generation of $89.7B in operating cash flow and $33.7B in net cash eliminates that dependence and breaks the historical capitulation pattern Burry's model assumes will repeat.

If the demand window extends to 2031, the marginal question is who can fill it. Counterpoint Research's Q2 2026 data put SK Hynix at ~50% of global HBM revenue (down from 64% a year earlier), Samsung Electronics at roughly 33% and Micron Technology at ~18% — but the more important column is bit-volume, where Micron's HBM4 yield ramp and calendar-2027 contracting window give it disproportionate leverage to the next two quarters. SK Hynix' M15X conversion capex runs ~$13–15B and2026 group capex is guided to the high-40 trillion won range (~$30B); Samsung's HBM3E qualification only landed in September 2025 and its HBM4 mass production started in January 2026.

Short-term, the path of least resistance is up: Q1 FY2027 revenue guidance of $61.5B ±$1.5B is $4.5B above consensus at $57B, and the 87.0% Q4 FY2026 non-GAAP gross margin implies ~$53B of quarterly gross profit. The first repricing event will be fiscal Q2 FY2027 earnings in March 2027, where Micron Technology reports the first quarter of calendar-2027 HBM bit shipments already contracted at higher year-over-year prices. Burry's likely tactical response — either a public close of the position or a roll to a longer-dated put — is itself a catalyst, since his July Substack disclosure added 4–6% of implied selling pressure that has now evaporated.