Position Status: +11.5% YTD | Conviction: 5→4/10 (downward) | Holding Decision: REDUCE 30-40% on next 2% rally --- 1. IS THE ORIGINAL THESIS INTACT? Short answer: The pricing power thesis holds. The margin resilience thesis is breaking. My original conviction stood on three pillars. Two are wobbling. - Pricing power in data center: Still demonstrable, but narrowing - ROIC resilience: Deteriorating faster than I expected - Margin expansion as competitive moat: This is where the thesis is cracking Here's what the data actually shows. What's working: Q1 2026 (10-Q filed 2026-05-20) shows $259.5B in assets vs. $206.8B in Jan 2026 — capex intensity is *real* and competitors can't match it . EPS dilution against massive revenue growth ($60B→$75B+ annualized run rate) remains modest — which *should* confirm pricing power is holding. But here's the problem: EPS dilation from $4.90 (Q4 2025) to $2.39 (Q1 2026) signals either one-time charges or margin compression I haven't fully accounted for. Neither answer is clean. What's breaking: My June 9 analysis flagged it directly: custom silicon can't scale fast enough to offset margin pressure. That wasn't speculation — it's now the *actual* constraint, not demand. When Apple, Meta, and Google are all building in-house silicon specifically to reduce NVIDIA dependency (as June 8 news confirms), the dynamic shifts in a specific and damaging way: unit ASP stays strong, but attach rates and total serviceable market flatten. That's margin compression by *volume*, not by price. Harder to see coming. Harder to reverse. --- 2. WHAT'S SURPRISED ME The insider selling is the real red flag — not the stock performance. Mark Stevens (Chief Financial Officer) sold 500,000 shares on June 4 . That's not routine rebalancing. That's a CFO cashing out heavily into new highs. When the finance guy exits at scale in a bull market, it typically means one of two things: 1. He sees margin risk the street hasn't priced yet — which is my concern 2. He's de-risking ahead of imminent earnings — the safer read, but the timing is odd given June earnings are right around the corner What makes this harder to dismiss: both the CFO *and* President are selling within days of each other. That's not opportunistic. That looks coordinated. **What I

Research these companies