SELL

• Parabolic extension 64% above the 200-day moving average
• Peak operating margins north of 80%, gross margins at 84% in most recent quarter
• Receivables growing faster than revenue
• Construction in progress more than doubled year over year
• Capex accelerating from $2.9B per quarter to $7.8B per quarter
• Shooting star rejection at upper Bollinger Band on elevated volume
• If earnings disappoint: gap down through 50 SMA at $935 to lower Bollinger Band at $880
• If earnings beat: dead cat bounce to distribution zone at $1041 where supply overwhelms demand
• Forward EPS of $155 assumes HBM cycle continues for 12 months but supply coming online in late 2026/2027 will normalize pricing power
• Conservative analyst models normalized EPS at $75-80 (vs aggressive analyst's 35% haircut), putting fair multiple at 12-13x even after 50% price decline
• Technical rejection at $1041 swing high, upper Bollinger Band at $1020, and failure to close above $1000 psychological level
• ATR of $54 means 6% daily volatility range
• 93% probability of no Fed cuts in 2026
• Beta of 2.22 means 11% MU decline in a 5% market drawdown
• Goldman upgrade narrative largely priced in; remaining flow from late buyers who get hurt
• Proposes 40-50% trim at current levels, secondary trim at $1041, hard stop at 50 SMA $935 on remaining 25-30%
• Could argue for 60-70% trim, keeping only 15-20% as true tail hedge

→ vs neutral: HOLD position is the most dangerous stance; 'waiting for data' is intellectually tidy but operationally suicidal; by the time September earnings confirm the trend, the stock will have gapped through the 50 SMA to the lower Bollinger Band

→ vs aggressive: 25-30% residual position is not 'insurance' but voluntary exposure to a binary event; insurance means sizing to the downside you are insuring against, not leaving room to re-enter at a lower level

→ vs aggressive: The 35% EPS haircut scenario is too conservative; forward EPS should be modeled at $75-80 in a normalized state given capex acceleration and incoming supply in late 2026/2027

→ vs aggressive: Position sizing does not account for ATR of $54, which is a 6% daily range and constitutes a margin of safety violation

Trim 30-40% at current levels, retain 40-50% as core position, hedge residual with deep OTM puts at $880-$900 strike

• MACD histogram crossed back positive after July-August negative territory; 10 EMA crossed above 50 SMA in early September - confirmed medium-term bullish crossover
• MA stack fully bullishly aligned with price above all three moving averages; 200 SMA rising from $372 in June to $610 today
• Shooting star candle is one signal, not a confirmed top - the second attempt to break $1,020, not the fifteenth
• HBM demand is structurally different from phone memory cycles due to multi-year hyperscaler capex commitments from Microsoft, Google, Meta, and Amazon
• AR grew 4.2x while revenue grew 3.7x - 13.5% relative divergence over multiple quarters of hypergrowth, not a smoking gun
• Inventory held flat at $8.5 billion - absence of inventory build contradicts end-demand softening thesis
• Capex-doubling is industry-wide (SK Hynix up 7%, Samsung doing same) for HBM3E/HBM4 capacity racing
• Beta of 2.22 and ATR of $54 makes $935 hard stop equivalent to a coin flip on routine volatility
• Residual 40-50% exposure paired with OTM puts defines maximum loss while preserving upside participation
• Tiered framework: first warning at daily close below 10 EMA ($965), second warning below 50 SMA ($935), full exit below lower Bollinger Band at $880 with elevated volume

→ vs aggressive: Shooting star thesis is one candle doing too much load-bearing; MACD and moving average crossovers confirm medium-term bullish structure, not bearish reversal

→ vs aggressive: Phone memory cycle framework is a category error when applied to HBM infrastructure with contracted hyperscaler demand - AI capex is being raised, not cut

→ vs aggressive: AR drift argument requires both AR growth AND inventory build to be a smoking gun; inventory held flat at $8.5B, so bear case has only one of two needed signals

→ vs conservative: $935 hard stop will get triggered by routine noise given beta of 2.22 and ATR of $54 - calling this 'discipline' is intellectual hygiene rather than actual risk management

→ vs conservative: Capex-doubling-is-sell-signal ignores that SK Hynix and Samsung are doing the same thing for HBM3E/HBM4 due to contracted hyperscaler demand, not cyclical expansion

→ vs conservative: Trimming 70% at $1,000 means 40%+ underperformance if MU breaks out above $1,041 on HBM guidance strength, then chasing higher to re-enter - a directional bet on cycle breaking dressed up as risk management

→ vs conservative: $935 stop conflates medium-term trend break with full exit signal; a daily close below lower Bollinger Band at $880 is the actual full exit trigger