Financebotresearch desk研究台

WDAY › analyze

WDAY · Analyze from before

ACCUMULATE Technology

Date: 2026-08-27 | Price: $193.57 (close) / $194.10 (after-hours) | Market Cap: $47.81B

1. What this updates

Baseline: Output/Stocks/Technology/WDAY/analyze-2026-04-08.md (full /analyze, 2026-04-08, price $119, BUY, conviction 8.0/10, fair value $179–240). Second-newest file, health-2026-03-20.md (price $136), was read in full and its own flags (SBC, aggressive FY2026 capital deployment, liquidity drawdown) are carried into this ledger as prior-pass evidence, not re-derived.

Event list since 2026-04-08 (this is what drove the re-test, not the calendar): 1. Q1 FY27 print (~May 21) — revenue +13.5%, backlog decelerating. 2. Reuters reports Silver Lake in take-private talks (2026-08-13) — stock +18–25% intraday, market cap briefly >$50B, 52-week high $249.85 set shortly after. Not present in the baseline at all. No signed deal as of this report. 3. Q2 FY27 print (today, 2026-08-27) — beat on revenue/EPS, guide raised modestly, total backlog growth missed analyst estimates ($27.4B actual vs. ~$28.6B est.) and decelerated sharply. Stock initially fell intraday on the miss, then closed +1.5%. 4. Analyst rating mix flipped from unanimous-bullish to genuinely split: Morgan Stanley (Underweight, PT $145, 7/21) and CLSA (Underperform, PT $92, 7/20) both initiated/downgraded before the M&A rumor, on fundamentals — this predates and is independent of the buyout speculation. 5. David Duffield (co-founder, >10% beneficial owner) sold ~107,500 shares roughly every 3–5 trading days, continuously, from before the baseline through 2026-07-10 (no sales recorded since — plausibly a blackout ahead of deal talks/earnings, not necessarily bearish). 6. A second one-time tax item distorted GAAP EPS again ($305M benefit from an intra-entity IP transfer in Q2 FY27) — the same mechanical class of trap the baseline flagged for FY2024's $1.058B deferred-tax reversal.

2. The delta ledger

Vendor-data checks run this pass (Phase 2), both confirmed live: - pitfall-yahoo-share-count-dual-class-fpi fires again on WDAY. Yahoo's raw sharesOutstanding = 201.00M; marketCap ÷ price implies 246.97M; the Q2 FY27 press release's own diluted count is 246.3M — confirms the vendor field, not the reconciliation, is wrong. Every per-share number below uses 246.3M, not the raw field. - pitfall-vendor-forward-eps-is-the-wrong-fiscal-year fires on WDAY's own forward P/E. Yahoo's forwardPE (15.31x) prices epsForward ($12.64) — which is FY2028, seventeen months out. The true current-year (FY2027, ends Jan-2027) figure sits in a different field: epsCurrentYear $10.76, priceEpsCurrentYear 17.99x — cross-checked bottom-up against H1 FY27 actual non-GAAP EPS ($2.66 + $2.75 = $5.41) plus a plausible H2 under the raised 31.0% margin guide, which lands within a few cents of $10.76. The baseline's "9.6x forward non-GAAP earnings" at $119 was very likely making the same wrong-year error — $119 ÷ $12.54 (its stated "FY2027E" figure) = 9.5x, and that $12.54 reads as a FY2028-shaped number, not a FY2027 one. This is not asserted as certain (the baseline's exact vendor pull wasn't logged), but it is flagged because the mechanism is identical and the arithmetic lines up. - pitfall-tax-valuation-allowance-round-trip-breaks-eps (a documented recurring WDAY-class trap) fires a second time: Q2 FY27 GAAP diluted EPS $2.57 includes a $305M one-time tax benefit (intra-entity IP transfer, $374M new DTA). Ex-item, normalized net income ≈$327M, EPS ≈$1.33 — in line with the trend, not the headline.

# Claim (as written in baseline) Type Status What moved it
17 Zero sell ratings / unanimous-bullish analyst consensus State ❌ RETRACTED Now 1 sell + 2 strong sell of 41; bullish% 60%→51%. Two bear calls (Morgan Stanley UW $145, CLSA Underperform $92) predate the M&A rumor and rest on fundamentals, not deal speculation. Primary-source (Yahoo ratings feed), single independent check, high confidence.
23 FCF-based fair value $142–240 (12x–20x FCF) Price 🔄 SUPERSEDED Re-derived from scratch on true share count and TTM FCF (see §6). New range $139–231, similar shape but the price has moved from deep-bear-case to base-case territory.
24 "Historically cheap" — P/FCF 11.3x, forward P/E 9.6x, FCF yield 8.9% Price 🔄 SUPERSEDED True TTM P/FCF is 16.8x (not 11.3x — stale shares + stale FCF). True current-year forward P/E is 17.99x (not 9.6x — wrong fiscal year, see above). Most of the "historically cheap" framing has closed via price appreciation (+63% since baseline), not because it was wrong then.
26 Verdict: BUY, conviction 8.0 Judgment 🔄 SUPERSEDED See §4 — new verdict ACCUMULATE, conviction 6.5.
27 Entry zone $117-135 Price 🔄 SUPERSEDED 30-40% below current spot; classic [[pitfall-stale-entry-zone-suppresses-a-name]]. Re-derived to $155-175 (§6).
15 Backlog (+17%) outpaces revenue (+13%) → demand isn't the problem, deal terms are just longer State 📉 DRIFTED This is the central finding of this pass. Total subscription backlog growth decelerated 17.0% (Q4 FY26) → 10.9% (Q1 FY27) → 8.0% (Q2 FY27) — two consecutive quarters, now running below subscription revenue growth (13.9%), and missed consensus ($27.4B vs. ~$28.6B est.). The baseline's own falsifying condition — total backlog growth no longer outrunning revenue — has now happened. 12-month backlog (the nearer-dated slice) is decelerating more gently (17.6%→15.5%→14.2%) and still sits above revenue growth, so near-term demand hasn't broken; it's the long-dated tail of bookings that's weakening. Breaks the thesis outright if total-backlog growth falls further below high-single-digits while revenue growth holds ~13% for another quarter.
8 FCF 3yr CAGR 29% Trend 📉 DRIFTED TTM FCF (Q3 FY26–Q2 FY27, reconstructed): $2.844B, only +2.4% over FY2026's $2.777B — a sharp deceleration from the 27-29% run-rate. The most recent quarter alone: Q2 FY27 FCF $460M vs. Q2 FY26 $588M = -21.8% YoY; OCF -15.6% YoY. One quarter, not yet a trend-break, but the direction is the wrong one. Breaks if Q3 FY27 FCF also prints down YoY.
12 SBC as % of revenue improving (20.9%→17.0%) Trend 📉 DRIFTED Q2 FY27 SBC/revenue = 17.44% vs. Q2 FY26's 16.69% — the first YoY uptick in a multi-year improving series. Also: capex reversed its multi-year decline, TTM capex now $238M vs. FY2026's $162M (+47%), likely AI/data-center infrastructure spend. Neither is alarming alone; both are the first counter-trend data points in metrics the baseline cited as structurally improving.
16 Balance sheet sound, net debt manageable, Debt/Assets 21% State 📉 DRIFTED Debt/Assets ticked to 23.65% (Q1 FY27 balance sheet, most recent complete). Cash + investments continued falling — $5.44B (FY26-end) → $4.35B (Q1 FY27) → $3.40B (Q2 FY27, press release). H1 FY27 buybacks alone ($2.9B) already equal all of FY2026's. This confirms rather than contradicts the 2026-03-20 health report's own flag ("not sustainable at this rate... another year like this would stress the balance sheet") — that prior caution is validated, not overturned.
21 No notable insider buying — "monitoring point" State 🔁 REFRESHED Old: soft, unquantified note. New: David Duffield sold ~107,500 shares roughly every 3-5 trading days, continuously, through 2026-07-10 (no sales recorded since, likely a blackout). Scale is material (>20 tranches, $12-21M each, in this window alone) but the cadence and round-lot size read as a pre-set 10b5-1 program, not a fresh bearish signal — flagged for visibility, not weighted as a red flag.
7 Capital allocation: aggressive buybacks + AI-suite M&A Structural 🔁 REFRESHED Continues and intensifies: $1.6B (Q1) + $1.3B (Q2) buybacks = $2.9B in H1 FY27 alone, matching the entire FY2026 total in half the time. True diluted share count fell 268.1M (FY26 avg) → 254.3M (Q1) → 246.3M (Q2), an 8.1% reduction in two quarters — the "buybacks now outpacing SBC dilution" inflection the baseline called is confirmed, more strongly than expected.
9, 11 Revenue deceleration (12-13% guide) / GAAP op margin expansion Trend 🔁 REFRESHED Revenue: Q1 +13.5%, Q2 +12.8%, FY27 guide raised slightly to $9.94-9.95B (13%). Landed almost exactly where the baseline expected. GAAP op margin: 11.8% (Q2), non-GAAP op margin 31.1% (Q2), FY guide raised 30.5%→31.0%. Both continuing in the direction and magnitude the baseline projected.
1, 2, 4, 5, 6, 10, 25 Platform/HCM+Finance integration moat · 97% gross retention · switching-cost economics · no-dividend capital-return model · competitive ranking (SAP > Microsoft/Oracle at the core, AI-native no threat at enterprise tier) · gross margin ~75-76% stable · evergreen rating 8/10 Structural/Trend ✅ CARRIED No contrary evidence surfaced on any of these. 97% retention re-confirmed for the Q2 FY27 period specifically (single independent source, consistent with the multi-year series — moderate confidence). Standing prior for structural claims applied per protocol; nothing here needed to move.
22 (Graham) Graham's Number void for SaaS Price ✅ CARRIED Recomputed: √(22.5 × 3.20 × 26.84) = $43.96 (fin.py, ties to hand calc). Same conclusion, same reasoning — not a usable anchor for a buyback-hollowed, SBC-heavy book.
3, 14, 19, 22 (Bogle) Market-share figures (33.8% core HR, ~50%+ large-enterprise) · ROIC trend · AI ARR $400M+/>100% growth · Bogle expected-return model — ⏳ UNTESTED Not independently re-pulled this pass. AI momentum has a partial proxy (AI now >25% of new ACV, >5,500 customers on organic agents, +35% QoQ) that is directionally supportive but does not confirm or refresh the specific $400M ARR figure. Flag for the next pass — this is the second report running these haven't been re-verified.

3. How the close calls were decided

Was the baseline's "backlog proves demand is fine" claim right, and is it now wrong, or was it always thin? The baseline's own falsification condition was explicit: backlog growth outrunning revenue growth meant demand wasn't the problem. That condition held at the baseline (backlog +17% vs. revenue +13%) and has now inverted (total backlog +8% vs. revenue +12.8%). This is not a single soft data point — it corroborates three ways: the total-backlog number itself, the consensus miss (analysts also expected faster backlog growth and didn't get it), and the two-quarter monotonic deceleration pattern (17.0%→10.9%→8.0%), which rules out a one-quarter lumpy-renewal explanation. Weighed against: the 12-month (near-dated) backlog slice is still outrunning revenue and decelerating more gently, so this reads as a genuine but partial break — DRIFTED, not RETRACTED, because near-term demand hasn't visibly cracked, only the long-dated tail.

Was the "historically cheap" valuation framing wrong at the time, or has it simply resolved? Both the share-count and forward-EPS vendor errors were almost certainly present in the baseline too (the mechanisms are dated to the same vendor, not to a change between April and August), but this pass cannot fully re-derive April's exact multiples without the historical payload. The honest call: the framing understated the multiple by a knowable, mechanical amount at the time, and the multiple has since re-rated further on top of that — both effects point the same direction (the stock is less cheap than the baseline stated, twice over). SUPERSEDED rather than RETRACTED, because the underlying $179-240 fair-value logic wasn't unreasonable, just built on a distorted "how cheap" starting point.

Does the balance-sheet drift and the FCF-quarter miss overturn the moat/structural case? No — weighed against six carried structural claims (retention, platform integration, switching costs, competitive ranking, capital-return model, evergreen rating) with no contrary evidence on any of them, two quarters of FCF/balance-sheet softness reads as a State-level wobble inside an intact model, not a Structural break. This is the "bad quarter inside an intact model" case the command flags explicitly — the Structural claims explain away the State weakness, they don't get overturned by it.

The Silver Lake speculation — Structural or Price? Per protocol, an M&A rumor is a Price force, not a Structural one, even though it currently dominates the stock's day-to-day behavior. It changes nothing about what Workday sells or how defensible it is; it changes what the market will pay and introduces event risk. It is treated that way in §6, not folded into the moat analysis.

4. Thesis persistence and conviction delta

Structural + Trend claims (14 total): CARRIED/REFRESHED = 10 (platform moat, retention, switching costs, capital-return model, competitive ranking, gross margin, capital-allocation identity, revenue deceleration, GAAP margin expansion, share-count/buyback inflection). DRIFTED = 2 (FCF trajectory, SBC%). UNTESTED = 2 (market share, ROIC).

Thesis persistence: 10/14 = 71%.

This is the textbook re-rating pattern the command flags: high persistence against a large price move ($119 → $193.57, +63%) — the business mostly held, the multiple did almost all the work. That argues for keeping this as an actionable name, not dropping it. It does not argue for the same conviction, because two independent things happened simultaneously: (1) the valuation cushion that justified an 8.0 largely closed on its own terms (§2, rows 23-24), and (2) a real, event-risk-shaped new variable (Silver Lake) now sits on top of the fundamental case and cannot be sized with the same tools as an FCF multiple.

Conviction: 8.0 → 6.5. Driven by: - Down: rows 23/24/27 (valuation cushion mostly closed), row 17 (consensus no longer unanimous — two independent fundamentals-based bear calls exist), row 15 (backlog thesis partially inverted), rows 8/12/16 (three DRIFTED State/Trend rows in the same direction — FCF, SBC%, balance-sheet cushion — is a pattern, not noise). - Held up by: 71% structural persistence, confirmed buyback-outpacing-dilution inflection (row 7), retention holding at 97%. - Not driving the number either way: the Silver Lake talks. A binary, unpriceable event with no signed terms is not something this framework should convert into either upside conviction or downside caution — it is named explicitly in §6 instead so the entry/trim logic can be read independently of whether a deal happens.

5. What is genuinely new

  1. Silver Lake take-private talks (Reuters, 2026-08-13). Reported ongoing for "months" before breaking publicly. No signed agreement as of this report. Analyst chatter (not a company statement) puts a plausible deal price at $230-250/share (a ~25% premium to the pre-rumor price implies roughly $219). Co-founders Bhusri and Duffield together control 68% of the vote — no deal happens without them, which also means no deal happens against their interests, a governance fact worth knowing given Bhusri's return as CEO seven months before the talks became public. This is speculative background, not a claim to be adjudicated — it is carried into the valuation as an asymmetric, unpriceable option, not a fact to bet on.
  2. Total backlog growth decelerating faster than revenue growth for two consecutive quarters (row 15) — the single most important fundamental finding this pass, because it is the baseline's own stated falsification test firing.
  3. A second GAAP-EPS-distorting one-time tax item ($305M IP-transfer benefit, Q2 FY27) — same mechanical class as the FY2024 deferred-tax reversal the baseline already knew to normalize for. Confirms this is a recurring feature of WDAY's reporting, not a one-off.
  4. The vendor forward-EPS wrong-fiscal-year trap and the dual-class share-count trap both fire on WDAY specifically (not just analogically from other tickers) — both now logged against this name in the Playbook (see §2).

6. Updated verdict

Valuation — rebuilt from scratch on true share count (246.3M diluted) and TTM figures

FCF-based (primary model). TTM FCF (Q3 FY26–Q2 FY27, reconstructed from quarterly filings): $2.844B → FCF/share $11.55.

Multiple Fair value/share
12x (bear — total-backlog deceleration continues, FCF stays flat) $138.60
15x (base-bear) $173.25
17x (base) $196.35
20x (bull — backlog re-accelerates, AI ACV mix keeps climbing) $231.00

Cross-check on current-year (FY2027) non-GAAP earnings, using the corrected basis (epsCurrentYear $10.76, not the vendor's mislabeled "forward" $12.64): a 16-24x band for a large-cap SaaS platform decelerating to low-teens growth with a 31% non-GAAP operating margin gives $172-258. Both models converge on roughly the same range.

Fair value: $175-235, central estimate ~$200. Current price $193.57 sits inside this range, near the base case — materially different from the baseline, where $119 sat near the bear case. Most of the deep-value setup the baseline identified has closed through price appreciation; what remains is a fairly-valued-to-modestly-cheap quality compounder, plus an unpriced M&A option.

Graham's Number: $43.96 — still void for the same structural reasons (SBC-suppressed earnings, buyback-hollowed book value). Not weighted.

Entry / Trim

  • Entry $155-175. This is a genuine re-derivation (the old $117-135 zone was 30-40% below spot and stale per [[pitfall-stale-entry-zone-suppresses-a-name]]), set with a margin of safety under the $175 floor of the fair-value range. Notably, this band roughly coincides with where WDAY traded just before the Silver Lake rumor broke (~$175 on 2026-08-12) — i.e., it is approximately where the stock would round-trip to on fundamentals alone if the deal talks lapse with no agreement. That coincidence is a feature, not a coincidence to explain away: it means the entry zone is coherent whether or not a deal ever happens.
  • Trim 18x fwd. ⚠️ This multiple is deliberately calibrated to the basis site.py actually renders, not to the current-year (FY2027) EPS discussed above. site.py's trim_from_multiple() computes dollar = multiple × (price ÷ vendor forwardPE), and the vendor forwardPE field on this name is the FY2028 basis (epsForward $12.64) per the pitfall fired in §2. 18 × $12.64 = ≈$228, which is where I actually want the trim to bite (near the low end of the analyst deal-price chatter and the top of this pass's fair-value range). The equivalent judgment on the true current-year (FY2027, $10.76) basis is ~21.2x — stated here so the next analyst can tell the two apart. Verify the rendered dollar after the next python .mcp/site.py build; if it doesn't land near $228, the vendor field moved and the multiple needs to move with it.

Break triggers (thesis fails)

  • Total subscription backlog growth falls further below high-single-digits for a second straight quarter while revenue growth holds near 13% — confirms the demand-conversion story was wrong, not a timing artifact.
  • FCF prints down YoY again in Q3 FY27 (two consecutive quarters).
  • SBC/revenue prints above ~18% for FY2027 as a whole (the multi-year improvement genuinely reverses, not just wobbles for a quarter).
  • Silver Lake talks are confirmed terminated and the stock does not hold above the $155-175 entry band on the resulting de-rating — would indicate the fundamental floor is lower than this pass estimates.

Upgrade conditions (back toward 7.5-8.0)

  • Total backlog growth re-accelerates above revenue growth for two consecutive quarters.
  • A signed Silver Lake agreement at a price this is the only forum where that would matter for a standalone-fundamentals framework: a signed deal converts this from a compounding thesis into a risk-arb spread trade, which is out of scope for this playbook — flag for the user's decision, don't auto-adjudicate it here.
  • SBC/revenue resumes its decline below 17%.

A portfolio-specific passage was removed from the public build.

7. What this pass did NOT test

  • Market-share figures (33.8% core HR, ~50%+ large-enterprise cloud HCM) — carried forward unverified for a second consecutive pass. If a third pass still hasn't re-checked these, treat them as assertions with seniority, not evidence.
  • ROIC trend — not recomputed this pass; the baseline's cash-basis ~25.7% and GAAP ~7.1%/ tangible ~14.4% figures are carried forward unverified.
  • AI ARR $400M+ figure — no updated dollar figure obtained. The >25%-of-new-ACV metric is a different measurement (contribution mix, not a stock of ARR) and should not be read as confirming or refreatuning the $400M base.
  • Bogle expected-return model — not rebuilt in full this pass; qualitatively, the starting multiple is now higher (17.99x vs. a genuinely-9-10x current-year multiple would have been), so a full rebuild would likely show a materially lower expected 5yr annual return than the baseline's 20-28% figure. Flagged, not computed — don't quote a number that wasn't derived.
  • Single-source items: the 97% retention re-confirmation and the AI-ACV-mix stat both rest on one search-derived source each, not a primary filing pull this pass. Moderate, not high, confidence.
  • Silver Lake deal probability/pricing — deliberately not modeled. No credible way to assign odds to an unconfirmed, unpriced private negotiation from public information, and the valuation/entry/trim above are constructed to be coherent without needing to.

Key Data Sources