Salesforce, Inc.Full report →1 / 14
CRMNYSEThe short version

Salesforce, Inc.

Salesforce is the world's largest seller of customer-relationship software, down 59% from its 2024 peak on fear that AI agents erode seat-based subscriptions. This distills a fixed test of whether it fits one contrarian investor's system.

From a $368 all-time high in December 2024 to a $150 low in June 2026, then a partial bounce to $171 — a 59% fall driven by multiple compression, not an earnings cut.
$171
Share price (Jul 17, 2026)
$163B
Market capitalization
3.3%
Adjusted FCF yield
$41.5B
FY2026 revenue
SwipeScroll
As reported

A capital-light software franchise: $41.5B revenue, 78% gross margin, $14.4B free cash flow

FY2021 → FY2026as reported · $
Revenue$41.5B+10%
Gross margin77.7%+0.5pp
Operating margin20.1%+1.0pp
Net income$7.5B+20%
Free cash flow$14.4B+16%
Open the full statements →
Standardized statements, FY2021 to FY2026, as reported.
  • Recurring by design. Subscription and support is 95% of revenue, sold on 12–36 month contracts paid in advance. Free cash flow has risen every year for a decade, from $1.7B to $14.4B.
  • Mature-software economics. A 77.7% gross margin and a 34.7% free-cash-flow margin on capex of 1.4% of revenue. The plant is code and data centers, not factories.
The fit

Does not fit the framework (P1 not met); contested: P2, P5

Not met
Year-10 durability gate (P1)binary gate; nothing offsets it
3.3%
Adjusted FCF yield vs 8–9% bar≈470 bps short
0.62
Prob. impairment is temporary (P5)contested · spread 0.26
Low
Confidence tier
  • The gate decides. Year-10 conviction needs a monopoly, a regulator, or heavy capital as the moat. Salesforce has none — capex is 1.4% of revenue and its own filing calls the market fragmented with low barriers to entry. Any proper doubt resolves the binary gate the same way.
  • The counter-fact, in the same breath. Revenue and free cash flow rose in all ten of the last ten years, and it is the #1 CRM vendor 13 years running. Rising history plus one unresolved AI doubt is exactly what the gate resolves against.
  • Two open questions. Free-cash-flow consistency (P2) and the temporary-versus-permanent diagnosis (P5) both came back contested across independent model families — reported honestly, not rounded to a clean answer.
The business

No single product dominates: six clouds sold to one shared customer base

FY2026 revenue by offering ($B)
Service, Sales and Platform are roughly a quarter each; data, marketing and services fill the balance.
  • System-of-record. Salesforce runs enterprise sales and service workflows, wired in deep enough that switching is slow and disruptive. That is the moat — real, but switching costs, not a regulated barrier.
  • Grown by acquisition. Slack, Tableau, MuleSoft and the $8.1B Informatica deal built the non-core lines. Revenue is spread across tens of thousands of customers, none more than 10% in any of the last three years.
  • US-centric. The Americas are 66% of revenue, Europe 24%, all of Asia Pacific 10%; 82% of assets sit in the United States.
The dislocation

Down 59% in 18 months — a multiple that halved, not an earnings cut

Path from the December 2024 high to the June 2026 low; the rebound to $171 is modest.
  • The multiple did the work. The forward earnings multiple compressed from ~33x to ~12x. Over the same span the FY2027 EPS estimate rose 7% while the price fell 34% — the numbers went up as the stock went down.
  • Not the framework's trigger. The canonical setup is an earnings cut the market anchors to. Here every recent quarter beat and consensus rose, so this is a de-rating on decelerating growth and AI-disruption fear, not a guidance shock.
Capitulation

Volume spiked to 3.5x normal at the bottom — capitulation, not orderly selling

3.5×
Peak 20-day volume vs pre-drawdown median
55.4M
Heaviest session (Jun 18, 2026)10.6× median, four days before the low
−59%
Peak-to-trough fall
  • Fired at the bottom. Volume ran an ordinary 6–8M shares through the first, drifting leg, then built to 17M a month at the June 2026 trough. Emotion-driven selling clustered into the final leg and exhausted at the low — the capitulation signature the framework wants.
  • The marginal buyer was the company. Salesforce bought back $12.6B of its own stock through the fall, with management calling the tape 'not a rational market.' The counter: reported growth did slow to ~9–10%, so some sellers were repricing a genuinely slower business.
Damage math

The cash engine grew through the fall — the usual numerator is roughly zero

Free cash flow — actual then consensus ($B)
FY2024–FY2026 reported; FY2027–FY2029 consensus. FCF rose 52% over two years, inside the price fall.
  • Price down, value up. Market cap fell ~$196B (−55%) from the peak while free cash flow rose 16% in the last year and forward consensus held flat-to-higher. The near-term cash hit the framework looks for is not negative — it is slightly positive.
  • So the de-rating reprices durability, not earnings. A reverse-DCF at a 9% discount rate says today's price embeds roughly zero perpetual FCF growth; the 2024 peak embedded about 5.5%. Five points of assumed long-run growth were stripped out.
The diagnosis

The panel split 0.62 on whether the reset is temporary or permanent

Ten-year DCF, off FY2026 FCF at a 9% discount rate
ScenarioEquity valuevs today
Temporary re-acceleration$413B+153%
Moderate maturation$265B+63%
Bearish (flat terminal)$172B+5%
Terminal decay (AI)$137B−16%
Today's $163B market cap sits between the two pessimistic rows.
  • For temporary. Pure multiple compression: ~29x trailing FCF to ~11x while FCF grew. Consensus never impaired the NPV, Q1 FY2027 EPS beat by 24%, and management lifted the buyback to $50B calling prices low.
  • For permanent. The 20%+ compounding engine has reset to high-single digits; recent acceleration leaned on Informatica and ~2 points of FX, with Agentforce Apps up only 9%. A reset terminal growth rate lowers NPV even with cash flow intact.
  • The ruling. Three blind judges put the probability the impairment is temporary at 0.62 (seats 0.42–0.68, spread 0.26). It leans temporary but does not settle — the permanent case commands a real minority.
Yield vs the bar

Cheap on reported cash flow, short of the bar on the framework's adjusted basis

FCF yield on the $163B market cap
Adjusted FCF = reported less stock-based comp less the five-year average of acquisition spend.
  • The adjustment is the whole story. Reported $14.4B FCF becomes ~$5.4B after ~$3.5B stock comp and a ~$5.5B five-year M&A charge for a serial acquirer — a 3.3% yield, roughly 470 bps under the 8% bar. Even charging nothing for deals leaves 6.7%.
  • Consensus agrees on the cash, not the adjusted yield. Forward reported FCF yield reaches ~10.9% by FY2029, but net of stock comp and acquisitions the adjusted forward yield is ~5.1%. Reversion above 8% within three years runs 20–25%.
Durability

Growth stepped down from mid-20s to single digits — the fact the market reprices

Revenue growth, year-over-year (%)
Revenue still rose every year — $8.4B to $41.5B over the decade — but the rate reset.
  • Deceleration, not decline. Consecutive-decline years on the framework's test are zero, so the structural-decline exclusion does not trip. But the level reset from ~25% to ~9–10%, and the mature core slowed (Sales +8%, Service +8%, Marketing & Commerce +3%).
  • Why the gate still fails. A higher year-10 is likely on base rates, but the framework wants very-high conviction from structure. Switching costs are execution-adjacent, and the 10-K concedes AI 'may disrupt our offerings.' Likely is not the standard the gate sets.
Self-help

A real $28B buyback, but stock comp eats two-thirds of it

Shares outstanding (millions)
The count climbed on stock-funded deals and comp, then fell three years running as buybacks overtook issuance.
  • 1.4% a year, not 10%. Gross repurchases of 116M shares over three years netted a 41M reduction (997M → 956M) — the rest offset stock comp. At a 10% adjusted yield the framework's flywheel retires ~10% a year; this delivers a fraction of that.
  • Willing, but sharing the till. A $50B authorization and a $25B accelerated buyback at low prices show intent; the counter is an $8B acquisition and a record buyback in the same year. The absurdity check is benign: ~11 years of FCF to retire the float, not the ~3 that flags an impossible price.
The clock

Its one comparable de-rate took 28 months to round-trip

Salesforce drawdowns of 30%+ since 2004
EpisodeDepthRound-trip
2008–09 (GFC)−71%18 mo
2015–16−34%6 mo
2020 (COVID)−36%5 mo
2021–22 (rate + de-rate)−59%28 mo
2024–26 (current)−59%not yet
The one comparable company-specific de-rate, 2021–22, took ~28 months to repair.
  • Repair by printed quarter, not one event. The buyback compounds per-share value quarterly, consensus already models a forward FCF yield clearing 9–10%, and growth is guided back to double digits by 2H FY2027 — the ingredients for the multiple to stop compressing on an 18-month clock.
  • But the base rate cautions. This name has never round-tripped a 55%+ company-specific de-rate inside 18 months, and the low was set about a month before this reading. Partial re-recognition over roughly two years is more supportable than a fast snapback.
The arithmetic

What a re-rating would take, and why the target price stays unstated

~0%
Perpetual FCF growth embedded at today's pricevs ~5.5% at the 2024 peak
$13B
Adjusted FCF needed to reach the 8% barvs ~$5.4B today
20–25%
Odds adjusted yield clears 8% within 3 years
$245
Sell-side mean target vs $17152 analysts, ~40% above spot
  • The formal re-rating price is unavailable. The tally leaves it null: stock-based comp was missing from the deterministic feed, so the normalized adjusted-FCF basis it needs could not be computed. The framework forbids improvising the figure.
  • What can be said. The street never left — 40 of 52 analysts at buy, targets ~40% above spot — so there is little upgrade fuel; a re-rating has to come from the buy side re-engaging with printed evidence. Consensus forward cash flow independently clears the yield bar, the shape the framework prefers even where the name does not clear the gate.
What to watch

A durable, cheap-looking cash machine that clears the fear test but not the year-10 gate

This is a distillation of a fixed, pillar-by-pillar fit test built tab by tab — evidence, not a recommendation.

Compiled from the full report · 2026-07-22 · For information, not investment advice.