When: Sequential decline in PP&E growth AND FCF dollars inflecting positive YoY

Then: RE-ENTER (fundamental thesis restoration)

SELL (partial trim 25-30% of position)

• Backlog of $678 billion is contracted revenue requiring delivery, not a valuation floor; provides visibility but not certainty against multiple compression
• FCF declined 24% from $25.7 billion (Q3 2025) to $19.6 billion (Q2 2026); capex deployment is risk, not reward, until returns materialize
• Buyback reduction from $7.4 billion to $4.6 billion per quarter ($2.8 billion reduction, $11 billion annualized) signals management's reduced confidence in equity valuation
• MACD crossed bearish with histogram expanding to negative 3.36; RSI cooled from 79 to 64; price pulling back from $505 swing high
• Downside to $420-450 is anchored by structural levels: 50 SMA at $421, July 30 gap-fill at $450, breakout retest at $463
• Upside to $530 faces resistance at $505 and $520 before prior swing zone, not an autopilot path
• Three binary catalysts in 5 trading days: Nvidia (8/26) priced for perfection with 20% downside risk on miss, Jackson Hole (8/28) with 70% hawkish odds, Anthropic IPO repricing risk
• PEG of 0.67 assumes expectations materialize; if they don't, multiple compresses rather than holds at 20.7
• Quality fundamentals are real (ROE 34%, net debt $19B vs equity $442B) but don't immunize against drawdowns or multiple compression

→ vs aggressive: Backlog ($678B) treated as valuation floor when it's a quality signal affecting the base revenue, not the multiple applied to it; backlogs shrink under economic stress

→ vs aggressive: FCF decline of 24% ($25.7B to $19.6B) dismissed as 'moat building' but represents real reduction in cash buffer, optionality, and shareholder returns

→ vs aggressive: Buyback reduction from $7.4B to $4.6B ($2.8B quarterly) is management's vote of confidence getting smaller, a signal risk managers should heed

→ vs aggressive: Short-term bearish technical signals (MACD bearish cross, RSI 64, $505 pullback) dismissed as 'within an uptrend' is dangerous framing; risk management responds to deterioration before collapse

→ vs aggressive: $530 upside framed as 'clear path' ignores $505 and $520 resistance; markets don't retrace to prior highs on autopilot post-drawdown to $355

→ vs aggressive: Recommendation to hold full position through 8/26-8/28 catalyst cluster in name 12% from 52-week high is reckless, not aggressive

BUY (trim 20% at $487, retain 80%, stop at $463, re-enter at $475-$480 or $505)

• Forward PEG of 0.67 vs trailing PEG of 1.57 is a real discrepancy—most institutional desks apply the trailing figure; quality at full price is still full price
• Nvidia report shows stock "priced for perfection" with 20% downside risk on miss; Polymarket has Warsh "Challenge" at 70% odds—these are defined risk events requiring defined sizing responses
• FCF dropped 24% while CapEx is $35.8 billion—reinvestment only creates value if returns materialize; FCF compression is a real cost, not a non-event
• Anthropic IPO odds at 36% (down 7.4 percentage points week-over-week) means that risk is fading, not building—conservative analyst is pricing maximum pessimism across all three events
• Current price $487 is middle of consolidation range; trimming at $487 is trimming into neutral, not into strength
• RSI has fallen from 79 to 64, MACD has crossed bearish with expanding negative histogram—momentum cooling
• Sell target at $495-$500 is flagged as resistance where momentum stalled; no guarantee those levels get tested before Jackson Hole
• $450 stop is ~3 ATRs below entry ($12 ATR on $487 stock implies $475-$499 daily range); $463 is the July 31 breakout retest and standard two ATR stop
• 20% trim reduces exposure meaningfully through August 26-28 catalyst window while maintaining participation if MSFT retests $505 toward $520-$530
• Trim should be framed as pause creating dry powder for re-entry at better levels, not as full exit

→ vs aggressive: Cherry-picked forward PEG of 0.67 when trailing PEG is 1.57—most institutional desks apply trailing figure; calling stock "cheap" at 66% gain over five years and within 3% of DCF fair value is a stretch

→ vs aggressive: Dismissed Nvidia and Jackson Hole as "volatility events, not trend-changers" despite data showing 20% downside risk on Nvidia miss and 70% Polymarket odds on Warsh "Challenge"—that's denial, not conviction

→ vs aggressive: Dismissing 24% FCF drop as reinvestment "non-event" ignores that reinvestment only creates value if returns materialize; requires AI monetization to land—it's a probability bet, not certainty

→ vs aggressive: Treating trim as leaving the position misses the point—trim creates dry powder for re-entry at better levels within the same trade

→ vs conservative: Overweighting near-term by treating three binary catalysts as if all three will go wrong simultaneously—baseline probabilities are mixed, not uniformly bearish

→ vs conservative: Selling into $495-$500 resistance levels that technical report flags as where momentum stalled; no guarantee those levels get tested before Jackson Hole, and Nvidia miss could cause gap down Thursday morning

→ vs conservative: $450 stop is too tight relative to ATR—$450 is July 30 gap-fill and probably overshoots where you'd actually want to buy back; stop logic should match re-entry framework