Dislocation

Dislocation

Salesforce fell 59.2%, from an all-time high of $367.87 on December 4, 2024 to $150.12 on June 22, 2026, and trades at $170.77 today — a real, deep drawdown. But its anatomy inverts the framework's usual trigger. No earnings cut caused it: the last three quarters beat consensus by 13.6%, 24.9% and 24.1%, and forward estimates rose while the price fell. This is a multiple de-rating on decelerating growth and AI-disruption fear, not a Centene-style guidance shock.

The drawdown

Source: peak, trough, current and depth from the deterministic capitulation gauge (derived from daily price history). Feature: fit_features.capitulation_gauge.drawdown.

Source: month-end closes from daily price history, as reported. The intramonth extremes — the $367.87 peak (Dec 4, 2024) and the $150.12 trough (Jun 22, 2026) — are carried in the KPIs above. Feature: fit_features.capitulation_gauge.

The fall took 565 days and came in three distinct phases, not one shock. The de-rating drift (Dec 2024 to mid-2025): from $367.87 the stock slid to roughly $265 by mid-2025 — down about 28% — on ordinary volume of 6 to 8 million shares a day. Under the framework's own test, that first leg is drift, not a moment: no single event and no volume spike. The grind (mid-2025 to late 2025): a further step down toward $230, tracking each quarter's ~9% growth print. The capitulation (Dec 2025 to Jun 2026): the sharpest leg, from about $265 to the $150.12 trough — where the volume finally spiked (below). The rebound since is modest: at $170.77 the stock sits 53.6% below the peak and 13.8% above the trough.

The trigger

There is no single dated adverse event of the kind the framework hunts — no guidance cut, no earnings miss, no restatement. What repriced Salesforce is a sequence of growth-deceleration prints against a hardening AI-disruption fear, and the primary record shows the fundamentals moving the opposite way from the price.

The clearest dated marker is the fiscal-2025 results call on February 26, 2025. Management guided fiscal-2026 revenue to $40.5 billion to $40.9 billion — growth of just 7% to 8%, a sharp step down from the 24%+ of prior years — and set non-GAAP EPS guidance of $11.09 to $11.17, while explicitly assuming only a "modest contribution" from Agentforce in fiscal 2026 [1]. The number that mattered was the growth rate, not a cut to earnings.

The de-rating character shows most plainly in what did not stop it. On the September 3, 2025 call, Salesforce beat, raised the low end of full-year revenue guidance to $41.1 billion to $41.3 billion, and announced a $20 billion expansion of its buyback authorization [2]. The stock fell 4.9% the next day and kept sliding. Fiscal 2026 closed at $41.5 billion of revenue, up 10% [3]. Prints that would have lifted the stock in prior years now met selling.

The mechanism behind the fear is legible in the company's own risk disclosure: the market for enterprise applications is "highly competitive, rapidly evolving, fragmented and subject to changing technology, low barriers to entry" [4]. The market's worry through 2025 and 2026 was that agentic AI erodes the pricing power of seat-based enterprise software — a "SaaSpocalypse," in the phrase management itself used. The final leg is illustrative: after the May 27, 2026 fiscal-Q1 print beat EPS by 24% and the stock jumped to $209.60 on June 1, it collapsed 28% to the $150.12 trough over the next three weeks, a move driven by soft next-quarter revenue guidance and AI-disruption fear rather than any deterioration in reported results. The June 2026 intraday trigger itself post-dates the filing corpus and is not resolvable to a primary document here.

Whether that fear reflects real, lasting damage to intrinsic value — or a mispricing — is the temporary-versus-permanent question decided in the Damage Math trial and the durability evidence in Durability. This tab describes what happened, not what it means for value.

The fear gauge

Source: monthly average daily volume from daily price history, as reported. Pre-peak baseline (median daily volume over the 180 days before the peak) was 5.22 million shares. Feature: fit_features.capitulation_gauge.volume_spike.

The capitulation gauge measures the peak 20-day average volume in the peak-to-trough leg at 3.5x the pre-peak median of 5.22 million shares — a genuine spike, and it fired where the framework wants it: at the bottom, not the top. Through the first, drifting leg of the fall, volume ran near its ordinary 6-to-8-million-share pace. It then built steadily as the stock broke down — to a 15.7-million-share monthly average in February 2026 and 17.1 million in June 2026, the trough month. The single heaviest day of the entire drawdown was June 18, 2026, four sessions before the low, at 55.4 million shares — 10.6x the pre-peak median — on a down day. Heavy, emotion-driven selling clustered into the final leg and exhausted at the bottom, which is the capitulation signature rather than an orderly repricing.

Who was selling

Reported short interest for Salesforce is not available in the official/public position feeds for this run, so the short-selling level and change cannot be quantified here — a data gap. No forced or structural sellers (index exits, fund liquidations, large insider disposals) are disclosed in the corpus for this window.

What the evidence does support is that the selling was valuation-sensitive, not information-driven: it repriced the multiple while the underlying estimates rose (next section). The most concrete counter-party fact is on the buy side. Salesforce was the marginal buyer of its own stock throughout the fall — repurchasing $12.6 billion of shares in fiscal 2026 and taking its share count from 997 million (FY2023) to 956 million (FY2026). On the February 25, 2026 call, with the stock near $190, management raised the repurchase authorization to $50 billion, said it returned "more than $14 billion or 99% of our free cash flow to shareholders," and characterized the tape directly: "this obviously is not a rational market," these are "some low prices" [5]. The company's capital-allocation behavior — treated in full in Self-Help — reads as the opposite of a forced seller. The strongest fact against a purely irrational read is that reported revenue growth did decelerate to roughly 9-10%, so some of the sellers were repricing a genuinely slower-growing business, not only fleeing sentiment.

Estimates versus price

The framework's canonical setup is a price fall that outruns an estimate cut. Salesforce presents the extreme version: there was no estimate cut at all. As the stock fell, consensus for the current forward year rose.

Source: FY2027 consensus normalized EPS from CapIQ estimate momentum (data/sp/estimates.json); CRM closes from daily price history, as reported. Feature context: fit_features.consensus_forward_yield.

From January 22 to June 21, 2026, the fiscal-2027 EPS estimate rose 7.1% — from $13.20 to $14.13 — while the price fell 33.5%, from $228.09 to $151.78. Fiscal-2027 revenue consensus held essentially flat over the same span (about $46.1 billion). The entire move was compression of the multiple, not a downgrade of the numbers: against fiscal-2026 non-GAAP EPS guidance of roughly $11.13, the December-2024 peak of $367.87 implied about 33 times forward earnings; at $170.77 against the $14.13 fiscal-2027 consensus, the multiple is about 12 times. The forward earnings base grew as the multiple more than halved.

Consensus also agrees on the cash side. Sell-side forward free-cash-flow estimates imply a yield on today's ~$163 billion market cap of 8.7% for fiscal 2026, 9.1% for fiscal 2027 and 10.0% for fiscal 2028 — the sell side is not modeling a collapse in cash generation; it is the buy side that repriced. Where those forward yields sit against the framework's balance-sheet-scaled bar is computed in Yield.

Bottom line

Salesforce is a genuine dislocation by magnitude — a 59.2% peak-to-trough fall with a real, late-firing volume capitulation. It is not the framework's canonical trigger: no earnings cut anchored the price. Estimates rose through the fall, every recent quarter beat, and the company itself bought stock hand-over-fist at what management called irrational prices. What repriced is the multiple — compressed from roughly 33x to about 12x forward earnings — on decelerating ~9-10% growth and the fear that agentic AI erodes seat-based software. Whether that fear is deserved is the trial's question, not this tab's.