When: Post-NVDA clarity Aug 27+

Then: Initiate or add 2-3% position

HOLD with explicit risk controls: hard stop at $249, no adding until confirmed close above $272 with positive MACD, pullback to $238, or post-Nvidia clarity; position sizing at 3-5% max (leaning 3% given negative MACD and binary event); immediate reassessment if AMZN gaps down more than 5% on Nvidia reaction.

• P/E of 26-29x on $2.83T market cap company is premium, not discounted; forward P/E signaling 17% earnings decline
• Forward EPS $10.39 vs TTM EPS $12.44 indicates earnings power deterioration
• Long-term debt up 123% YoY; issued $59.5B new debt in single quarter; Q1 2026 FCF negative $18.2B; TTM FCF only $3.2B on $775B revenue
• Current ratio 1.03 indicates tight liquidity; D&A rising 30%+ YoY as capex assets come online
• Anthropic IPO probability 82-91% is not certainty; $15.68B gain already monetized in Q2 2026, limiting residual upside from residual stake
• AMZN near 52-week highs with negative MACD, declining 10 EMA, and binary Nvidia event 48 hours away creates meaningful left tail risk
• With beta of 1.45, meaningful market drawdown could produce 10-20% downside from $262; downside is not capped
• NVDA miss/drag scenario: AMZN could fall 5-10% in a day vs 3-5% upside if NVDA beats—negative expected value trade entering fresh
• Macro signals: Gold at $4,640, silver up 18% MoM (risk-off hedges); Polymarket 86% no Fed cuts (rates higher for longer, growth multiples pressured)
• Smart money 13F signals are 2-month-old snapshots (June 30, 2026) with no visibility since; consensus positioning suggests being late, not early

→ vs aggressive: 26-29x P/E is not a 'decade-low valuation' or discount; it is a full premium multiple on a company with declining earnings power and a 17% forward earnings decline signaled

→ vs aggressive: Capex buildup is framed as 'strength' but actually represents largest debt-funded infrastructure buildout in Amazon's history; long-term debt +123% YoY and $59.5B debt issuance in one quarter is balance sheet degradation, not financial strength

→ vs aggressive: 13F smart money signals are 2-month-old snapshots with no post-June 30 visibility; even if positions intact, consensus mega-cap positioning means being late, not early

→ vs aggressive: HOLD is not 'asymmetric to upside'; with beta 1.45 and AMZN near 52-week highs on negative MACD, downside is 10-20% and fully symmetric, not capped

→ vs aggressive: NVDA probability-weighted expected value math does not support adding on dip; NVDA miss drags AMZN 5-10% vs 3-5% upside if beats—that is negative expected value, not positive

→ vs aggressive: HOLD preserves capital through binary event but requires explicit stop at $249, not just inaction; conviction-driven inaction differs from risk-managed patience

HOLD

• HOLD is correct: when aggressive and conservative camps both land on the same action for different reasons, it's typically a robust signal
• Position size at 2-3% is the appropriate risk control; don't tighten artificially or expand
• Stop at $249 is fine because it's anchored to the 50 SMA, a meaningful technical level; don't tighten or widen
• Three entry triggers are well-calibrated: pullback to $238 with RSI below 35, breakout above $287 on volume with positive MACD, or post-Nvidia clarity
• Valuation is fairly valued at 26-29x normalized P/E—not cheap (doesn't support adding) and not expensive
• Anthropic catalyst is probabilistic with 82-91% probability and 9-18% tail risk; reason to maintain exposure, not add
• Capex story has a 3-quarter lag before payoff is known; insufficient information to be bullish or bearish
• 13F filings are mildly informative and stale, not a central pillar
• Balance sheet leverage (123% debt increase) is operating capex into highest-return segment, not financial engineering
• Amazon is fairly valued at $2.83 trillion with forward earnings projected to decline 17% YoY; entry at $262 neither cheap nor expensive
• Time-box the HOLD: soft review date around September 5 if no directional resolution after Nvidia

→ vs aggressive: Cannot redefine HOLD as the aggressive stance; that's rhetorical sleight of hand. HOLD means no action. Aggressive would be adding or tightening, not preserving exposure.

→ vs aggressive: "HOLD is asymmetric to the upside" is mathematically false; asymmetry requires capped downside. The position has full exposure to a 5-10% gap down if Nvidia disappoints. That's directional, not asymmetric.

→ vs aggressive: Valuation argument half-right: 26-29x normalized P/E is not "decade-low distressed"—it's a fair multiple for a quality compounder with above-average uncertainty. Anchoring on Motley Fool narrative is not math.

→ vs aggressive: Treating 82-91% Anthropic probability as certainty is wrong; there is a 9-18% tail risk of delay or disappointment.

→ vs aggressive: Over-weighting two 13F filings as a buy signal; these are mildly informative and stale.

→ vs conservative: Over-engineering risk management on an already-small position. The 2-3% size IS the risk control. Tighter stops on a small position get whipsawed out on noise.

→ vs conservative: Capex framing too pessimistic: balance sheet change is operating capex into high-return segments during demand inflection (AWS 16%, Pinterest $4B through 2031, Marvell silicon), not financial engineering.

→ vs conservative: Assuming Anthropic catalyst is fully priced in is wrong; Q2 monetization of $15.68B was on a portion of the stake; residual stake plus Bedrock strategic validation is a separate repricing event.

→ vs conservative: Dismissing 13F filings entirely is equally wrong as overweighting them.