AI compute demand growing exponentially. Jevons Paradox: as AI inference costs drop, total compute consumption increases faster than efficiency gains. Memory (HBM/DRAM) is the bottleneck — Micron (MU) is the primary beneficiary.

Bull case: MU breaks ATH → semiconductor cycle expansion continues → options leverage generates 3-5x returns → portfolio approaches $150K+.

Bear case: Semiconductor correction, AI capex pullback, or macro shock. MU $420 = stop level. Trailing stops active on all positions.

Edge: 15 autonomous scanners, multi-language signal processing, 3-model LLM Council consensus, hard-coded risk controls. System rotated from oil to compute in 48 hours when thesis exhausted.

Phase 1 (Mar 9-31): Oil/Geopolitical thesis — GUSH $45c +332%, OXY calls +331%

Phase 2 (Apr 1-25): AI Compute rotation — MU $440c +196%, portfolio to $37K

Phase 3 (Apr 25+): Semiconductor expansion — MU $620c +288%, NVDA, FCX added → $132K (+501%)

• Max 40% concentration per ticker (shares + options combined)

• Only BUY — no shorts, no naked options, no margin ever

• 6 hard-coded risk controls in risk_gate() — non-overridable

• -20% drawdown = automatic kill switch

• $3,000 daily loss limit halts trading

• Trailing stops: MU 8%, SOXL 10%, NVDA 8%

• LLM Council (3 models) required for trades >$1K