When: Q3 2026 beat-and-raise with HashiCorp contribution disclosure and no goodwill impairment

Then: Full benchmark-weight re-engagement

HOLD

• Requires confirmed close above $245 with volume before adding position
• Together AI deal economics unverified—no recognized revenue, cluster live Q1 2027 only
• FCF yield of ~5.4% ($12B trailing FCF) is a value trap if multiple compresses from 21x to 15x
• $74.6 billion goodwill on balance sheet creates impairment risk; Q1 2026 already showed $445M restructuring charge
• $10.5 billion HashiCorp acquisition creates integration cycle risk through back half of 2026
• Q3 earnings in mid-October represent binary event risk with precedent for 25% single-session drawdown
• 200-day SMA at $264.68 acting as resistance magnet on weak volume
• 86% probability of zero Fed rate cuts in 2026 constrains macro tailwind
• Current entry at $236 with potential downside to $211 on earnings miss (11% capital loss)
• Dividend yield 2.84%—requires 4 years to recover one quarter's capital impairment
• Stock still trading below both 50-day and 200-day moving averages
• July 14, 2026 low of $211 not structurally established support—just a remembered level on panic print

→ vs aggressive: Confirmation above $245 with volume is not 'chasing'—it is validation that buyers have absorbed supply overhead; July 14 proves trading without it is gambling

→ vs aggressive: Nine outlets covering the Together AI press release is not nine independent catalysts—it is one announcement syndicated across nine channels; counting headlines is not measuring conviction

→ vs aggressive: FCF yield only matters if the multiple holds; a 5.4% yield on a stock that just crashed 25% is not a bargain—it is irrelevant if the multiple compresses to 15x on a revisit to July lows

→ vs aggressive: Goodwill is not a '25-year-old accounting fear'—it is a real, recurring, multi-billion-dollar non-cash charge; shareholders lost real money in the last goodwill cycle

→ vs aggressive: Diversification does not protect against a single print moving the stock 25%—July 14 happened precisely because the market didn't know the catalyst, not because it did

→ vs aggressive: The 'opportunity cost' math is wrong—waiting forgoes one year of return, but entering now risks one quarter of capital impairment; these are not symmetric outcomes

→ vs aggressive: $211 as a 'structurally established' floor is incorrect—a level tested once on extreme volume from panic selling is a remembered level, not structural support; markets retest such levels regularly

HOLD with conditional partial scale-in (max 50% of standard unit) on 10 EMA support confirmation at $232 with RSI above 45; full stop at $211 with wick tolerance to $208-$209; no additions beyond partial position until Q3 earnings

• July 14 capitulation print was 67 million shares - exhaustion point, not structural floor; wicks to $211 get retested when sellers return
• Bounce from $211 occurred on declining volume with no fundamental driver - not buyers defending a level
• Together AI deal provides credible reference customer for enterprise AI inference, changing institutional narrative even if not transformational
• 5.4% FCF yield cushions downside - if 15% drawdown occurs, still earning ~8% annually in dividend plus FCF return
• August 12 volume of 4.37 million shares is ambiguous - low volume near key levels means either exhaustion or compression before re-test
• IBM has multiple segments (Red Hat, HashiCorp, consulting, mainframe cycle, watsonx, quantum) - Q3 miss would not replicate July 14's 25% drawdown
• Q3 earnings will be telegraphed through guidance and channel checks; magnitude of move will be smaller than July 14
• Probability-weighted return of adding now: 30% probability of 15% drawdown vs 70% probability of flat-to-modest move makes current addition negative EV
• 2.84% dividend yield cushions time preference; no profit-taking needed if not adding aggressively
• 10 EMA at $232 with RSI above 45 and volume below post-July 14 average represents defensive entry criteria

→ vs aggressive: Calls $211 'structurally established' based on single panic print - weak logic; exhaustion points get retested when original sellers exit at breakeven or stop-losses trigger second cascade

→ vs aggressive: 'Sellers are exhausted' claim based on 4.37 million share August 12 volume is cherry-picked interpretation; low volume near key levels means either exhaustion OR compression before re-test - data doesn't force bullish reading

→ vs aggressive: Frames waiting for $245 as forfeiture of 8% annual return, but dividend yield is not risk-free when 25% of principal can be lost in single session; probability-weighted EV of adding now is negative

→ vs aggressive: FCF yield is not a free option - capital is at real risk until trend confirms

→ vs conservative: Dismisses Together AI deal as 'syndicated press release' - strategically incomplete; Reuters, Barrons, and institutional press framing as legitimate neocloud validation reflects how smart money models IBM optionality

→ vs conservative: Too dismissive on FCF yield - 5.4% yield is irrelevant only if downside doesn't matter; it provides real cushion that changes risk calculus

→ vs conservative: Treats Q3 earnings as binary 25% drawdown risk - ignores IBM's multi-product diversification (Red Hat, HashiCorp, consulting, mainframe, watsonx, quantum); Q3 will have telegraphed catalyst unlike July 14

→ vs conservative: Entry trigger at $245 reclaim is too rigid - forces 4% chase after 25% breakdown; more nuanced criteria allow capturing recovery without overcommitting