IdeaAI capex cycle signal
→
Research3-week deep dive
→
ModelBull $210 / Base $180
→
IC Vote3-0 Approved
→
EntryAvg $148, 8% NAV
→
Now$178, 12.5% NAV
Alex Chen — Sizing & approval
Sarah Lin — Research lead
Priya Patel — Factor validation
David Park — Risk sign-off
Mike Wang — VWAP execution
Thesis: Hyperscaler AI capex cycles typically run 3–5 years. NVDA's CUDA moat and H100/H200 supply constraints create persistent pricing power. At entry, trading at 25x forward EV/EBITDA vs software comps at 40x — significant discount for the infrastructure layer of AI. Key risk: AMD/Intel competitive response and potential customer in-house chip programs (Google TPU, Amazon Trainium). Monitoring: Quarterly datacenter revenue growth and new customer concentration.