ServiceNow NOW

Reference table

1. What the business is, and the two things the filings had to tell us

ServiceNow describes itself as delivering cloud-based solutions to automate business services, built on the "Now Platform". Chief executive William R. McDermott; approximately 29,187 employees per the vendor profile; Santa Clara, California; NYSE-listed since 2012.

Revenue disaggregates two ways, and the vendor supplies both — with one structural change worth noting.

By type (10-Q, three months to 30 June 2026):

Subscription is 97.2% of revenue. That is as clean a recurring-revenue profile as exists at this scale.

By product line (seg_prod) — and the structure changed in fiscal 2025:

Every year reconciles to reported revenue exactly. But ServiceNow stopped disaggregating Digital Workflow from IT Operations Management in fiscal 2025, collapsing them into a single "License and Service" line. That is a genuine reduction in disclosure granularity — the ITOM line was growing 22% in fiscal 2024 and is now invisible — and it is stated here as a finding rather than passed over.

By geography (seg_geo):

All years sum to reported revenue exactly, and the North American home market is present at 62.9% — the SYK/DHR failure mode does not occur. Geographic mix is remarkably stable, which for a company growing 21-24% a year means all three regions are compounding at similar rates.

The forward-looking disclosures — which are the ones that matter

From the 10-Q filed 2026-07-23:

> "As of June 30, 2026, our RPO was $29.0 billion, of which 46% represented cRPO. RPO and cRPO both increased by 21% compared to June 30, 2025."

> "Our renewal rate was 98% for each of the three and six months ended June 30, 2026 and 2025."

> "Subscription services arrangements typically have a three-year duration, and we have experienced a renewal rate of 98% over the last three years."

Those three sentences are the strongest evidence in this file against the bear case, and they should be read carefully for what they do and do not say. RPO of $29.0 billion is 1.97x trailing revenue — nearly two years of contracted work already signed. cRPO of approximately $13.3 billion converts inside twelve months, which is roughly 90% of the FY2026 consensus revenue of $16.2 billion. A 98% renewal rate on three-year contracts means the customer base is not leaving.

What they do not say is anything about renewals in 2029. The bear case in Section 7 is explicitly a claim about what enterprises do when their current three-year contracts expire, having by then had two years of cheap agents. RPO and renewal rate are lagging measures of decisions already taken. They are the right numbers to watch and they will warn late.

2. The $8.8 billion the payload does not contain

This is the highest-yield finding in this dive and it comes entirely from the 10-Q.

> "On April 20, 2026, we acquired all outstanding shares of Armis Security Ltd. ('Armis'), a cyber-exposure management and cyber-physical security solutions provider, for preliminary purchase price consideration of approximately $7.6 billion, settled in cash."

> "On March 2, 2026, we acquired all outstanding shares of Veza Technologies, Inc. ('Veza'), a privately held AI identity security company ... for approximately $1.2 billion, substantially in cash."

The purchase-price allocations, from the 10-Q:

Armis's intangibles break into developed technology $1,950M (5-6 year life), customer relationships $473M, order backlog $54M and brand assets $53M. Goodwill is not deductible for income tax purposes.

And a third, earlier: Moveworks, Inc., acquired 15 December 2025 for $2.4 billion — an enterprise-search and virtual-agent business, which explains the jump in goodwill from $1,273M to $3,578M in the fiscal 2025 balance sheet.

The financing, from the 8-K of 2026-05-15 and the 10-Q's debt note:

The May 2026 offering was $4,000,000,000 of aggregate principal, and the 10-Q states the proceeds "were used to fund the repayment of outstanding borrowings under the Term Loan" — so Armis was bridged with a term loan in April and termed out in May.

Note the coupons. ServiceNow's pre-existing debt was $1.5 billion at 1.40%. Its new debt is $4.0 billion at 4.25% to 6.30%. The thirty-year tranche at 6.30% is the market's price for ServiceNow's terminal value, put in writing by bond investors in May 2026 — and it is a considerably less enthusiastic number than the equity market has historically paid.

What the vendor payload contains about any of this: nothing. bal_a ends at 31 December 2025, before Veza and before Armis. cf_a ends at the same date and shows acquisitionsNet of −$1,084M, which is Moveworks and Logik.io — not the $8.8 billion of 2026. No field in quote, profile, km_ttm, ratios_ttm or est reflects any of it.

The balance-sheet consequence, using the 10-Q:

Read the last three rows together. ServiceNow added $6,076M of liabilities and its equity went down $448M, because it simultaneously bought back stock: treasury shares rose from 18,498 thousand to 38,593 thousand, and treasury stock at cost from $3,045M to $5,371M — $2,326 million of repurchases in six months. The company borrowed $6 billion and bought back $2.3 billion of its own stock into a 35% drawdown. Whether that is conviction or leverage-on-leverage is a legitimate difference of opinion; it is certainly not passive.

3. The margin collapse, and what caused it

The June-quarter income statement, from the 10-Q:

The 10-Q explains the cost line directly:

> "Cost of subscription revenues increased by $405 million ... primarily due to increased headcount and increased costs to support the growth of our subscription offerings including costs to support customers in regulated markets. Personnel-related costs ... increased by $115 million ... Depreciation expense related to infrastructure hardware equipment and expenses associated with software, maintenance and other costs, which together support the expansion of data center capacity, increased by $63 million ... Expenses associated with our contractual commitments with third-party cloud service providers increased by $63 million ... In addition, amortization of intangible assets increased by $153 million ... as a result of acquisitions."

So the $405 million increase in subscription cost decomposes as roughly $153M acquisition amortisation, $115M personnel, $63M data centre, $63M third-party cloud, and $11M residual. The acquisition amortisation alone is 3.8 points of the 6.8-point gross-margin decline — more than half. That portion is non-cash and will lap; the data-centre and cloud portions are cash and will not.

Three further observations that a reader must have.

First, net income is flattered. "Other income (expense), net" was +$206 million against −$3 million a year earlier. The 10-Q attributes this to "net upward adjustments" on strategic investments — non-marketable private equity holdings, now $2,073M on the balance sheet, marked up on observable price changes in later financing rounds. That is a $209 million swing on a $298 million net income. Strip it and the quarter's net income would have been roughly $135M rather than $298M. These are real gains under the accounting, but they are mark-to-model on private companies and they will not recur reliably.

Second, interest income fell 39.7% ($70M against $116M) because the securities portfolio was spent on Armis. Meanwhile interest expense is rising as the $4.0 billion of new notes accrues. That reversal — from earning interest to paying it — is worth roughly $200 million a year at current balances and rates and has barely begun to appear in the reported numbers.

Third, stock-based compensation is enormous and is excluded from every multiple anyone quotes. The 10-Q's supplementary table gives June-quarter stock compensation of $96M + $13M + $179M + $283M + $84M = $655 million, on revenue of $3,987M — 16.4% of revenue in the quarter, and stockBasedCompensationToRevenueTTM reads 14.83%. Fiscal 2025 stock compensation was $1,955M. The 23.5x FY2027 forward multiple in Section 6 is computed on a non-GAAP number that excludes all of it. The mitigating fact, and it is a genuine one: buybacks of $2,326M in six months more than offset the dilution, and shares outstanding actually fell 1.3%.

Annual history, for context:

Two things stand out. Revenue has compounded 24.0% a year for five years with no down year — through a rate shock, a software recession and an AI panic. And GAAP operating margin had been expanding steadily, from 4.4% to 13.7%, until the acquisitions arrived. The 4.1% printed in the June quarter is not a return to the old level; it is a new, acquisition-driven distortion on top of a genuinely improved cost structure. That distinction matters and it is why the margin collapse is a quality-of-earnings issue rather than a demand issue.

4. The enterprise-value rebuild

Vendor: km_ttm.enterpriseValueTTM = $128,090,456,680, being market capitalisation of $122,140,456,680 plus exactly $5,950,000,000 of net debt.

We reverse-engineered that $5,950M and it decomposes precisely as: total debt $7,517M + operating lease liabilities $936M ($114M current + $822M non-current) − cash and equivalents $2,503M = $5,950M.

Two errors in one figure. The vendor adds $936M of operating lease liabilities to debt and omits $4,204M of marketable securities — $2,161M current plus $2,043M long-term — which the 10-Q explicitly counts as liquidity:

> "When assessing sources of liquidity, we also include cash and cash equivalents, marketable securities and long-term marketable securities totaling $6.7 billion as of June 30, 2026."

The rebuild:

> Market cap $122.140B (1,033,862 thousand shares × $118.14)

> + total debt $7.517B

> − cash and equivalents $2.503B

> − marketable securities, current $2.161B

> − long-term marketable securities $2.043B

> = corrected enterprise value $122.950B, on net debt of $810M

The overstatement is $5.14 billion, or 4.2% — modest against a $122 billion equity value, but it changes the leverage characterisation from "1.7x levered" to "effectively unlevered", and that matters when the whole debate is about whether this company can survive a demand shock. If one further counted the $2,073M of strategic investments, ServiceNow would be net cash by $1,263M — but those are non-marketable private holdings and we exclude them, noting their existence.

The direction of travel is nonetheless real and adverse. At 31 December 2025 ServiceNow had net cash of $1,323M. At 30 June 2026 it has net debt of $810M — a $2.1 billion swing, plus $3.3 billion of securities consumed, in six months. A reader relying on bal_a sees the December picture and would conclude the company is unlevered. It no longer is, and the 6.30% coupon on the 2056 notes is what that cost.

5. Cash flow and capital returns

Free cash flow has compounded 28.2% a year since fiscal 2022 and the margin has widened 450 basis points to 34.5%. That is genuinely excellent and is the strongest number in this file. incomeQualityTTM of 3.178 — operating cash flow more than three times net income — reflects both deferred revenue growth and the fact that GAAP net income is depressed by non-cash stock compensation and amortisation.

Stock compensation is falling as a percentage of revenue — 19.3% to 14.7% over four years — which is the right direction, but the absolute number is still $1,955 million a year. The key ratio is buyback against stock compensation, and it has just turned favourable: $2,326M of repurchases in six months against roughly $1,105M of stock compensation, and share count fell 1.3% from 1,047.3M to 1,033.9M. For most of ServiceNow's history the buyback merely slowed dilution; it is now genuinely shrinking the count.

No dividend. dividendPerShareTTM is zero and correctly reported.

Capital expenditure of $868M in fiscal 2025 is 6.5% of revenue (capexToRevenueTTM 4.94% on the trailing base), which is high for a software company and reflects owned data-centre infrastructure — the 10-Q names "expansion of data center capacity" and "contractual commitments with third-party cloud service providers" as growing cost lines.

6. Valuation — priced in or room?

At $118.14 (market cap $122.14B, 1,033,862 thousand shares, corrected enterprise value $122.95B):

Estimate coverage is excellent through FY2027 and thins after: 31-35 analysts on FY2026 and FY2027, 13 on FY2028 EPS, 11 on FY2029 and 5 on FY2030. The FY2030 EPS row is excluded from every conclusion.

The FY2028 EPS dispersion is the most informative single datum in this section, and it is extraordinary: low $3.633, mean $6.096, high $9.139. The high is 2.52 times the low. For comparison, Fortinet's equivalent spread in this batch is 1.49x and Parker-Hannifin's 1.18x. That 2.5:1 range is the AI-disruption debate expressed as a number, and it is why no fair value for this company should be presented with false precision.

est.ebitdaAvg and est.ebitAvg are fixed-ratio fabrications and are REJECTED in full. ebitdaAvg is exactly 35.17% of revenueAvg in every year from FY2024 through FY2030; ebitAvg is exactly 31.77%. ServiceNow's actual fiscal 2025 EBITDA margin was 22.6% and GAAP operating margin 13.7%. The vendor's FY2025 ebitdaAvg of $4,657M overstates the reported $2,999M by 55%. All forward valuation runs on epsAvg and revenueAvg.

Note also that the FY2023 estimate row contains netIncomeAvg: 0 and epsAvg: 0 — a null row that would produce an infinite multiple if used. It is discarded.

est.epsAvg validates. The FY2026 figure of $4.072 reconciles against the first half's reported non-GAAP actuals ($0.97 + $0.90 = $1.87) plus the Q3 estimate of $1.03, implying a Q4 of $1.17 — consistent with ServiceNow's disclosed fourth-quarter seasonality. The EPS row carries the entire forward valuation and is reliable.

Peer context. The vendor peer set contains genuine comparables for once — Salesforce, Intuit, Palo Alto Networks, AppLovin, Shopify — alongside semiconductor names (Applied Materials, Lam Research, Qualcomm), Arista and Uber. The relevant frame is Salesforce and the broader enterprise-application complex, which the knowledge base treats as a single trade. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn — but note that several knowledge-base claims group ServiceNow explicitly with Salesforce, Workday and Adobe, and one groups it with Snowflake, which is also in this batch.

6a. What today's price assumes (the inversion)

At $118.14 — 23.5x the FY2027 consensus and 19.4x FY2028 — the price embeds:

• Revenue reaches $19,245M in 2027 and $22,807M in 2028. (Consensus; 35 and 33 analysts.) That is +18.7% then +18.5%, decelerating from the +24.0% just delivered but still a doubling of revenue in roughly four years. The nearest check is the 2026-10-28 print at a consensus $4,095.6M, which would be +20.2% year on year.
• Non-GAAP EPS reaches $5.018 in 2027 and $6.096 in 2028 — +23.2% then +21.5%, faster than revenue, which requires the operating margin to keep expanding through the acquisition amortisation. (Consensus; 26 and only 13 analysts.)
• The renewal rate stays near 98% through the 2028-2029 contract cycle. (Our number; the current figure is filing-confirmed.) This is the assumption the entire valuation rests on, and it is the one the bear case attacks directly. A conviction-80 knowledge-base claim argues renewals get cut 50-75% as agents commoditise vertical software. If that is right, nothing in the 2026 or 2027 numbers warns you — the contracts are already signed.
• The $8.8 billion of security acquisitions earns its cost of capital, and $13,618M of goodwill and intangibles — 43% of total assets — is not impaired. (Filing-confirmed transactions; our assumption on the return.) The $2,530M of Armis intangibles amortise over five to six years, which is roughly $450 million a year of non-cash charge against GAAP earnings and nothing against the non-GAAP number the multiple uses.
• The market keeps paying 20-25x forward non-GAAP earnings for a business it used to pay 45-60x for. (Our number.) At 18x FY2028E the stock is $110; at 30x it is $183. The multiple, not the earnings, is the entire variance — and the multiple is a proxy for a belief about 2030.

6b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+21.5%, from FY2027E $5.018 to FY2028E $6.096) + multiple drift (COMPRESSION, from 23.5x to roughly 22x on the respective forward year, −6%) + shareholder yield (+1.9%, entirely buyback, net of stock compensation) ≈ +15% to +17%.

Our base assumes mild further compression, and the reason is honest uncertainty rather than a view. ServiceNow has already de-rated from 45-60x forward to 23.5x. We do not think we can forecast where a multiple that has halved on a philosophical argument settles, and applying further compression is the conservative treatment rather than an analytical claim. If the AI-tool-use thesis proves right and the debate resolves, this multiple does not stay at 23.5x — it re-rates hard, which is what the $200 bull case is.

Almost the entire expected return is earnings growth. If the multiple simply held at 23.5x on FY2028E, the price would be $143 (+21.2%); at 18x it would be $110 (−7.0%); at 30x, $183 (+54.8%). The spread between those is 62 percentage points, and it is not resolvable by analysis of the company. It is resolvable only by evidence about what enterprises renew in 2029.

6c. Variant perception (where we differ, what would surprise)

• We have NO variant perception against the street on value, and that is a verdict input. Our base of $135 sits within $3 of the consensus $138. When our own arithmetic lands on the crowd's number, we have no edge on price, and no-edge names default toward Watch. What we do have is a variant perception against the DATA: every screen shows ServiceNow with $1.3 billion of net cash and no acquisitions, and the filings show $810 million of net debt and $8.8 billion spent.
• We differ from the reported GAAP earnings in both directions and net roughly to zero. GAAP net income of $298M is overstated by $206M of strategic-investment mark-ups and understated by $153M of incremental non-cash acquisition amortisation. Both should be stripped: the underlying quarter is roughly $250-280M of net income on $3,987M of revenue, not $298M and not $135M. We say this because both distortions are large and they point opposite ways, and a reader who adjusts for only one will get a worse answer than one who adjusts for neither.
• We differ from the loudest bull in the knowledge base on his standing, not his argument. Twelve of the 45 claims naming ServiceNow come from Jensen Huang, whose company co-built an AI reasoning model with ServiceNow and sells hardware that "accelerates" its platform. His argument — that agents are tool users and therefore raise system-of-record usage — is a good argument and is corroborated independently. But a commercially interested party supplying 29% of the bullish evidence is exactly the concentration the attribution policy exists to catch, and we half-weight him and say so.
• Positive surprise that would force a re-rate: cRPO growth accelerating above 22% on 2026-10-28; the renewal rate holding at 98% for a fourth consecutive reported period; GAAP operating margin recovering above 8% as the amortisation is absorbed; or separate disclosure of Armis revenue showing clear accretion.
• Negative surprise that would break the thesis: the renewal rate falling below 96% — this is the single number that would confirm the bear case and it has not moved in a year; cRPO growth decelerating below 15%; an impairment against the $13,618M of goodwill and intangibles; guidance implying FY2027 revenue below $18.5 billion; or the $2,082M of short-term debt being refinanced at a materially higher rate than the 4.25-6.30% strip already issued.

Synthos fair values

All three anchors are multiples of the FY2027 consensus non-GAAP EPS distribution — mean $5.018, low $4.392, high $5.250 — which at 26 analysts is the best-covered forward year that is not essentially guidance. The FY2028 distribution (low $3.633, mean $6.096, high $9.139) is deliberately NOT used as the anchor, because a 2.5:1 low-to-high range would let us produce almost any answer we wanted. It is quoted as a cross-check and as evidence of uncertainty.

• Bear ~$80 — 18x the FY2027 consensus LOW of $4.392. Cross-check: 22.0x the FY2028 consensus LOW of $3.633; roughly 3.6% below the 52-week low of $83.00. The scenario: the agent-disruption thesis is right, the renewal rate breaks below 96% in 2028, cRPO growth decelerates into the low teens, the $13,618M of goodwill and intangibles takes an impairment, and a decelerating software company with new leverage earns a market multiple. −32.3%.
• Base ~$135 — 26.9x the FY2027 consensus MEAN of $5.018. Cross-check: 22.1x FY2028E of $6.096; corrected enterprise value of $140.5B on FY2027E revenue of $19,245M is 7.3x sales. Sensitivity, stated openly: 24x gives $120 — essentially spot — and 30x gives $151. The whole answer lives in a 24-30x band. The scenario: revenue and EPS land on consensus, the renewal rate holds, the acquisitions integrate without impairment, GAAP margins recover as amortisation is lapped, and the multiple stabilises somewhere between the panic level and the old level. +14.3%, plus roughly 1.9% of buyback yield.
• Bull ~$200 — 38x the FY2027 consensus HIGH of $5.250. Cross-check: 21.9x the FY2028 consensus HIGH of $9.139; 4.1% above the 52-week high of $192.23 and 15% below the street's high target of $236. The scenario: the tool-use thesis is validated in reported numbers — cRPO reaccelerates, the agentic platform monetises at a premium, the security acquisitions cross-sell into the installed base — and the market restores a growth multiple to a company still compounding revenue near 20%. +69.3%.

Base is 14.3% above spot; asymmetry roughly 2.15:1 to the upside (32.3% down, 69.3% up). That is the best base-case return and the second-best skew in this batch. What stops it being a Buy is that our base is the street's number, the knowledge-base lane is genuinely split with the freshest claim bearish, the entry is into a 28.5% three-month bounce, and the risk being taken is a terminal-value risk that no near-term evidence can retire.

7. Knowledge base — 45 claims, both sides of the argument, and a conflicted bull

This is the largest and most contested lane in this batch by an order of magnitude, and it is handled at length because the argument in it IS the investment case.

Raw entity hits: 45 unique (33 matching ServiceNow; 13 matching a case-sensitive standalone NOW; one overlaps both). Quarantined and excluded by policy: 3. Admissible: 42. Excluded from the conviction pool as conflicted-partner voice: 12. In the conviction pool: 30.

Context for that number: the median name in this batch has three claims. Parker-Hannifin has two distinct. Progressive has zero. ServiceNow has forty-five.

The conflicted bull, handled first

Twelve of the 42 admissible claims — 28.6% — come from a single channel attributed to Jensen Huang, chief executive of NVIDIA. He is not ServiceNow management, so the strict management-voice rule does not apply — but he is a commercially interested party, and the file proves it in his own claims:

> 2025-05-15 · bullish · conviction 75 · horizon: fact · entities: ServiceNow, Nvidia

> "ServiceNow and Nvidia jointly built [a reasoning model — model name redacted per house rules], a reasoning model that lets agents break down problems, research the web/PDFs, and return answers."

A chief executive whose company co-develops products with ServiceNow, sells hardware that "accelerates" its platform, and whose own valuation depends on the enterprise AI narrative is not an independent voice on whether enterprise software survives AI. All twelve of his claims are bullish, at convictions of 75 to 90, and all twelve make essentially the same argument:

> 2026-02-04 · bullish · conviction 85

> "The thesis that AI will replace software tool companies is 'the most illogical thing in the world'; AGI will use tools like these, not reinvent them."

> 2026-02-26 · bullish · conviction 80

> "Market wrongly fears a SaaS apocalypse; agents are tool users, so tool/system-of-record usage rises as enterprises deploy agents."

> 2025-05-15 · bullish · conviction 90

> "Every company needs AI agents curated by IT; ServiceNow already sits in every IT department, so it's destined to be the OS of enterprise AI agents."

Treatment: these twelve are excluded from the conviction pool, quoted here at half weight, and flagged. The argument is a good one and is independently corroborated (see the no_priors claims below). The point is that it should be weighed on the independent corroboration, not on twelve repetitions from an interested party.

CONCENTRATION SENSITIVITY TEST — mandatory, and run. Removing all twelve Huang claims takes the lane from 42 to 30. The remaining 30 split approximately 17 bullish, 11 bearish, 2 neutral. The bullish case survives without him — it is carried by no_priors, invest_like_the_best, empire, jacob_shapiro and Dan Ives — but the net conviction weakens materially, from clearly positive to roughly balanced, and the recency tilt goes negative. That is a real change in the reading and is why the test exists.

The bear case, from the conviction pool

> 2026-04-24 · BEARISH · conviction 80 · speaker: Chamath · channel: all_in · entities: CRM, NOW, SNOW, ADBE

> "Agents got so cheap enterprises spin up bespoke alternatives to vertical SaaS; unit cost is ~30% of value, so customers cut renewals 50-75% and cash flows become unpredictable."

> 2026-01-31 · BEARISH · conviction 80 · speaker: jordi_visser · speaker_role: independent · entities: CRM, MSFT, ADBE, NOW, SAP

> "AI ([a frontier assistant — lab and product names redacted per house rules] eating Excel/enterprise) is forcing high-multiple long-duration software names — Salesforce, Microsoft, ServiceNow, SAP, Adobe — to be rerated; the beginning of a massive rotation out of abundant software."

> 2026-07-14 · BEARISH · conviction 62 · speaker: Josh Brown · speaker_role: independent · entities: CRM, NOW, WDAY — THE MOST RECENT CLAIM IN THE LANE

> "IBM's pre-announcement was a warning shot: enterprise SaaS spending is not a corporate priority and probably won't be for the rest of the summer as capex shifts elsewhere."

> 2026-04-24 · bearish · conviction 45 · speaker: Adam Parker · channel: compound_and_friends · entity: NOW

> "ServiceNow's beat on revenue but miss on earnings with lower guidance margins is the worst bucket; the real repricing is ~4% on 2030 fundamentals, uncertain if a 2-3yr entry."

> 2026-04-10 · bearish · conviction 65 · channel: compound_and_friends · no named speaker

> "Software is in a bloodbath on AI-disruption fear; market ignores record earnings/estimates and prices a much lower terminal value years out."

> 2024-07-01 · bearish · conviction 58 · speaker: Gavin Baker · speaker_role: independent · entities: CRM, NOW, TEAM

> "Most application software may just go away, quickly, because AI startups replace human labor directly rather than tooling it."

The Adam Parker claim deserves particular attention because it is falsifiable and has partly been tested. It describes the pattern "beat on revenue but miss on earnings with lower guidance margins" as "the worst bucket". The June quarter, reported three months later, produced exactly that shape: revenue of $3,987M against a $3,927M consensus (a beat) and GAAP operating margin of 4.1% against 11.1% (a collapse). The claim was about the April print; the pattern repeated in July. That is a correctly specified, prospectively useful observation and it is weighted accordingly.

The bull case, from the conviction pool

> 2026-04-17 · conviction 85 · channel: no_priors · STANCE TAGGED "BEARISH" — SEE DEFECT NOTE BELOW · entity: ServiceNow

> "The 'SaaS apocalypse' is overstated — replicating an enterprise platform app with a language model costs ~10x more once you add GPU/token/rebuild costs."

> 2026-04-17 · bullish · conviction 80 · channel: no_priors · entity: ServiceNow

> "ServiceNow positions as the AI control tower integrating hyperscalers, LLMs, and 800+ systems of record — the connective fabric for agentic business."

> 2026-04-17 · bullish · conviction 70 · channel: no_priors · entities: ServiceNow, Armis

> "Cybercrime is a ~$1T/month problem; ServiceNow's Armis acquisition adds IT/OT and device visibility to build a full-landscape security purview."

> 2026-02-13 · bullish · conviction 82 · speaker: Dan Ives · speaker_role: independent · entities: CRM, NOW, ORCL, MSFT

> "Calling software structurally dead is the most misguided call of his career; Salesforce, ServiceNow, Oracle, Microsoft will be the hearts and lungs of the AI revolution, use cases hitting in 6-18 months."

> 2025-12-02 · bullish · conviction 88 · horizon: principle · channel: invest_like_the_best · entities: Salesforce, Workday, ServiceNow

> "The vast majority of market-cap creation goes to the market leader — true even in non-network-effect enterprise markets; being #2 or #3 is painful."

> 2026-01-09 · bullish · conviction 62 · channel: jacob_shapiro · speaker_role: independent · entity: NOW

> "Built a position in ServiceNow — a sophisticated software company applying AI/system integration inside S&P 500 businesses with very high retention; reasonable, not dirt-cheap, prices given growth and cash flow."

> 2026-03-03 · bullish · conviction 65 · channel: compound_and_friends · entities: ServiceNow, NOW

> "Added ServiceNow alongside the CEO buying more; also holds Salesforce through the AI-threat selloff."

The no_priors cluster is the most operationally specific evidence in the entire lane — the "10x cost" claim and the "800+ systems of record" claim are the only two in 45 that make a testable engineering argument rather than a market-structure one. They are also dated 17 April 2026, three days before the Armis close, and one of them anticipates the strategic logic of that acquisition.

A DATA-QUALITY DEFECT IN THE KNOWLEDGE BASE, reported. The claim "The 'SaaS apocalypse' is overstated" is tagged stance: bearish, conviction 85. Its content is unambiguously bullish for ServiceNow. The tagger has recorded the stance toward the apocalypse thesis rather than toward the company. We count it as bullish and flag the mis-tag, because a mechanical stance aggregation over this lane would score it backwards — and because if this error occurs once in 45 claims it presumably occurs elsewhere in the base.

The neutral middle, which is the honest reading

> 2026-02-07 · neutral · conviction 75 · channel: all_in · entities: Salesforce, Figma, ServiceNow, Adobe

> "Software sold off hard but revenue is stable/rising; drop reflects AI uncertainty discounting future cash flows, not deteriorating fundamentals."

> 2026-03-27 · bearish · conviction 65 · channel: all_in · entities: Snowflake, ServiceNow, Atlassian, Workday

> "SaaS names are being re-rated hard: market-cap/free-cash-flow payback periods have roughly halved as market prices in AI fragility of their cash flows."

Those two claims, from the same channel seven weeks apart, describe the situation more accurately than either extreme: fundamentals are fine, the multiple has halved, and the market is pricing a terminal-value risk rather than an earnings risk. That is precisely what the numbers in Sections 2 and 3 show.

Conclusion on the lane

The Synthos knowledge base has a great deal to say about ServiceNow and does not resolve the question. After excluding three quarantined records and half-weighting twelve claims from a commercially interested party, the pool of thirty splits roughly 17 bullish, 11 bearish, 2 neutral — but the bearish claims are fresher (January to July 2026 versus a bullish set skewed toward February to April), carry higher average conviction in the named-speaker subset, and include the single most recent claim in the entire lane. The bullish case rests on a genuinely good engineering argument (agents are tool users; rebuilding is 10x more expensive) and a genuinely good market-structure argument (leaders capture the market-cap creation). The bearish case rests on a claim about renewal behaviour in 2029 that no 2026 number can test.

Net conviction: MIXED-CONTESTED. Breadth 9 distinct channels. This lane raises our confidence that we understand the argument and does not raise our confidence about the answer, which is why it supports a Watch with a named flip level rather than a position.

8. Data integrity — what we rejected and why

Nine findings, and the first is the largest in this batch.

1. THE MAJOR FINDING — $8.8 billion of acquisitions and $6.0 billion of new debt appear in NO vendor field. The 10-Q filed 2026-07-23 discloses Armis Security Ltd., acquired 2026-04-20 for approximately $7.6 billion in cash ($5,323M goodwill, $2,530M intangibles) and Veza Technologies, Inc., acquired 2026-03-02 for approximately $1.2 billion ($826M goodwill, $356M intangibles), funded by a term loan subsequently refinanced with $4.0 billion of notes issued May 2026 at coupons of 4.25%, 4.70%, 5.05%, 5.40% and 6.30%, plus $2,082M of short-term debt. bal_a ends 2025-12-31 — before both transactions. cf_a ends at the same date and shows acquisitionsNet of −$1,084M, which is Moveworks and Logik.io, not the 2026 deals. Total debt went from $1,491M to $7,517M; goodwill and intangibles from $4,699M to $13,618M, 43% of total assets; and the company from $1,323M of net cash to $810M of net debt. Every screen a reader can consult shows the December picture. We present the June balance sheet from the filing and label the vendor's as stale.

2. km_ttm.enterpriseValueTTM is wrong in two directions — REJECTED and rebuilt. The vendor's $128,090,456,680 implies net debt of exactly $5,950,000,000, which decomposes as total debt $7,517M plus operating lease liabilities $936M minus cash $2,503M. It adds operating leases to debt AND omits $4,204M of marketable securities ($2,161M current, $2,043M long-term) — which the 10-Q explicitly counts as liquidity, stating "cash and cash equivalents, marketable securities and long-term marketable securities totaling $6.7 billion as of June 30, 2026." The vendor reports net debt of $5.950B; the filing supports $810M; we use $810M. Corrected enterprise value $122.950B against $128.090B; evToSalesTTM 8.69x → 8.35x; netDebtToEBITDATTM 1.705x → 0.23x.

3. est.ebitdaAvg and est.ebitAvg are fixed-ratio fabrications — REJECTED in full. ebitdaAvg is exactly 35.17% of revenueAvg in every year FY2024 through FY2030; ebitAvg is exactly 31.77%. ServiceNow's actual fiscal 2025 EBITDA margin was 22.6% and GAAP operating margin 13.7%; the vendor's FY2025 ebitdaAvg of $4,657M overstates the reported $2,999M by 55%. All forward valuation runs on epsAvg and revenueAvg.

4. The FY2023 estimate row is null and would produce an infinite multiple — DISCARDED. It reports netIncomeAvg: 0, epsAvg: 0, epsLow: 0 and epsHigh: 0 alongside a revenueAvg of $9,955M. Not used for anything.

5. returnOnEquityTTM is reported inconsistently across two blocks — both rejected. ratios_ttm.returnOnEquityTTM is null; km_ttm.returnOnEquityTTM is 0.1377. The two blocks disagree within the same file. Since equity fell $448M in six months while $8.8 billion of acquisitions were absorbed, neither figure is a stable measure. returnOnInvestedCapitalTTM of 5.5% is also distorted by the acquisitions and is quoted with that caveat. No return-on-capital measure in this file is currently reliable, and that is stated rather than papered over.

6. The 5-for-1 stock split is real, was applied correctly, and was verified — a clean finding. The 10-Q states: "On December 5, 2025, our board of directors approved and declared a 5-for-1 split of our common stock ... record date of December 16, 2025 and an effective date of December 17, 2025 ... All references made to common share, equity award and per share amounts ... have been retroactively adjusted." The vendor has applied it consistently: weightedAverageShsOutDil runs 1,012M to 1,050M across fiscal 2020 to fiscal 2025, all split-adjusted, and the implied share count from market capitalisation divided by price (1,033,861 thousand) matches the 10-Q cover figure of 1,033,862 thousand to within one thousand shares. This check was run because the price level demanded it, and it passed — no APP-style pre-and-post-basis mixing is present.

7. seg_prod lost a disaggregation level in fiscal 2025 — reported as a disclosure change, not a defect. Fiscal 2024 and earlier split "License and Service" into Digital Workflow Products and ITOM Products; fiscal 2025 reports a single combined line of $12,883M. All years still reconcile to reported revenue exactly. The change is the company's, not the vendor's, but it removes the ability to track IT Operations Management separately — a line that was growing 22% in fiscal 2024.

8. GAAP net income contains two large distortions pointing in opposite directions — both disclosed, neither corrected in the vendor file. The June quarter's $298M of net income includes +$206M of "other income (expense), net" from upward marks on non-marketable strategic investments (against −$3M a year earlier) and is depressed by $153M of incremental acquired-intangible amortisation in cost of revenue. A reader adjusting for one and not the other gets a worse answer than one adjusting for neither. The underlying figure is roughly $250-280M.

9. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — tech used. Quote: $194.726 / $81.24. Tech: $192.23 / $83.00. Intraday versus close. We use tech throughout. Note tech.pct_from_hi is −38.5% while tech.max_dd_from_peak is −49.5% — these differ correctly, because ServiceNow's six-year peak predates the last twelve months. Both are right; this is the legitimate case.

Not defects, correctly reported and independently confirmed: inc_q June-quarter revenue of $3,987M, operating income of $162M, income before tax of $438M, tax of $140M and net income of $298M all match the 10-Q income statement to the dollar; the March-quarter figures likewise; seg_geo sums to reported revenue exactly in every year with the 62.9% North American home market present; seg_prod sums exactly in every year; dividendPerShareTTM of zero is correct; stockBasedCompensationToRevenueTTM of 14.83% reconciles to $1,955M on $13,278M; incomeQualityTTM of 3.178 correctly reflects operating cash flow of $5,444M against net income of $1,748M; profile.currency is USD and inc_a[0].reportedCurrency is USD — no currency mixing; and earn_cal uses a consistent non-GAAP basis across all five reported rows, so the beat series is like-for-like.

Non-equity tripwire — checked and passed. NOW is common stock, $0.001 par value, NYSE-listed (10-Q balance sheet: 3,000,000 thousand shares authorised, 1,072,455 thousand issued, 1,033,862 thousand outstanding). The $118.14 price is not par-like; beta is 0.93; there is no dividend; volume was 17.83M shares (~$2.1B of turnover); and the 52-week band of $83.00 to $192.23 is a 132% range. This is common equity. The unusual share count relative to ServiceNow's history prompted the split check in finding 6, which passed.

9. Technicals

• Price $118.14, up 3.46% on the day. −38.5% from the 52-week high of $192.23; −49.5% from the six-year peak; +42.3% above the 52-week low of $83.00. Position within the annual range: 32.1st percentile.
• Moving averages, split: +11.2% above a 50-day of $106.21, which is rising; −4.9% below a 200-day of $124.27, which is falling. The intermediate trend has turned; the long-term trend has not.
• RSI 62.6 — firm but not overbought. MACD +2.72 — positive.
• Relative performance, and it has two distinct phases: 3-month +28.5% vs SPY +7.6% and QQQ +7.7% — sharp outperformance; 6-month +0.1% vs SPY +11.1% and QQQ +15.7% — flat against a rising market; 12-month −35.4% vs SPY +24.3% and QQQ +30.8%. ServiceNow underperformed the S&P 500 by 59.7 points over twelve months and outperformed it by 20.9 points over three. That is a violent de-rating followed by a violent bounce, inside the same year.
• Sentiment, and the dispersion is the story: 60 buy, 8 hold, 1 sell across 69 analysts — overwhelmingly positive. Consensus target $138.00 (+16.8%), median $132 (+11.7%), high $236 (+99.8%), low $85 (−28.1%). The high is 2.78 times the low — by far the widest target dispersion in this batch.

What the technical picture does to the entry. The 200-day average at $124.27, 5.2% overhead, is the meaningful level: reclaiming it would convert a bounce into a trend change. Below it, a 28.5% three-month move in a stock that fell 35% over twelve months is a rally within a downtrend until proven otherwise. The honest position is that buying today means paying up 28.5% from the low into an unresolved debate, and that the far better entries in this name over the last twelve months have all come after bad news rather than before it.

10. Insiders — a director grant cycle and one small sale

Six of the eight records are the annual non-employee director grant — five directors receiving identical 3,260-share awards at a zero price on 2026-05-21, plus one later grant. These carry no information.

The single discretionary transaction is a sale: Teresa Briggs, a director, sold 1,595 shares at $108.70 on 2026-05-28 — seven days after receiving a 3,260-share grant. That is roughly $173,000 and reads as ordinary tax-related liquidation of half a grant. It is a mild negative and no more than that.

No officer appears in the file, so no inference about management is available from this block. What is available is second-hand: a compound_and_friends claim dated 2026-03-03 states the position was added "alongside the CEO buying more", which would be a discretionary open-market purchase by William McDermott. That transaction is NOT in this file and we cannot verify it — it is reported as an unverified knowledge-base assertion, not as a fact.

The company itself is the meaningful buyer: $2,326 million of treasury purchases in the six months to 30 June 2026, at prices that averaged well below where the stock traded a year earlier.

11. Verdict, kill-criteria and flip conditions

Watch — and this is the closest call in the batch.

What is genuinely excellent: revenue compounding 24.0% a year for five years with no down year, and still growing 24.0% in the most recent quarter; subscription revenue at 97.2% of the total, growing 24.5%; $29.0 billion of remaining performance obligations, up 21%, with 46% converting inside a year; a 98% renewal rate that has not moved in a year on three-year contracts; free cash flow compounding 28.2% a year to $4,576M at a 34.5% margin; stock compensation falling from 19.3% to 14.7% of revenue while buybacks of $2,326M in six months shrank the share count 1.3%; and $6.7 billion of liquidity. And a price 38.5% below the 52-week high, 49.5% below the peak, at 23.5x forward against a history of 45-60x.

What holds us back: a base fair value of $135 that sits within $3 of the street's $138 — no variant perception on price; a knowledge-base lane of 45 claims that is genuinely split, whose freshest entry is bearish and whose loudest bull is a commercially interested partner supplying 29% of the evidence; GAAP operating margin of 4.1% and a 680 basis point gross-margin decline; net income flattered by $206M of private-company mark-ups; a balance sheet that moved from $1,323M of net cash to $810M of net debt on $8.8 billion of acquisitions and $6.0 billion of new borrowing, at coupons up to 6.30%; $13,618M of goodwill and intangibles now 43% of assets; $1,955M a year of stock compensation excluded from every multiple quoted; and an entry 28.5% above the three-month low.

The distinction that matters. We are not arguing that ServiceNow is impaired — the reported numbers say clearly that it is not. We are arguing that the question the price turns on is not answerable from the reported numbers, that our own valuation lands exactly on the crowd's, and that the marginal knowledge in this dive is about the balance sheet rather than the thesis: the market is debating 2030 while the company quietly borrowed six billion dollars and bought two security businesses that no screen reflects. Where we do have an edge here is on the data, not the view — and a data edge on a name where our price target equals consensus is a Watch, not a position.

Pre-registered KILL criteria — what would take this to Avoid:

• The renewal rate falling below 96%. It has been 98% for at least three years and is the single cleanest possible confirmation of the agent-disruption thesis.
• Current RPO growth decelerating below 15%, from the 21% just reported. cRPO leads revenue by roughly a year and is contracted, so a deceleration there is not a timing issue.
• Any impairment against the $13,618M of goodwill and intangibles, which would confirm that $8.8 billion was spent at the top.
• GAAP operating margin failing to recover above 8% by mid-2027, which would mean the acquisition drag is structural rather than amortisation timing.
• Guidance implying FY2027 revenue below $18.5 billion, against the $19.245 billion consensus.
• The $2,082M of short-term debt refinancing above 7%, which would say the credit market has moved further against the terminal value than the 6.30% thirty-year already implies.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• Price below approximately $100 (roughly 19.9x FY2027E, 16.4x FY2028E) with cRPO growth still above 18% and the renewal rate at 98%. At $100 the base case is +35% against a bear case only 20% lower, and the arithmetic becomes compelling rather than merely favourable. That is 15.4% below today's close and 20% above the 52-week low.
• cRPO growth ACCELERATING above 22% on 2026-10-28. This is the strongest available evidence that agentic demand is additive rather than substitutive, and it is the number the bear case has to break.
• GAAP operating margin recovering above 8% while revenue growth stays above 18%, which would show the acquisitions integrating rather than diluting.
• Separate disclosure of Armis revenue and growth, which would allow an organic calculation for the first time and remove the largest unmeasured item in this dive.
• A decisive shift in the knowledge-base lane — specifically, the bearish named-speaker claims being retracted or contradicted by their own authors, or a run of fresh independent bullish claims with the operational specificity the no_priors cluster showed.

Where NOW fits in the Synthos Framework Portfolio. The enterprise-software sleeve, at 0% today with a 2-2.5% target on a fill near $100 or on cRPO accelerating above 22%. Batch overlap note, and it is material: SNOW is in this batch and appears in the same knowledge-base claims as NOW — Chamath's conviction-80 bearish claim names "CRM, NOW, SNOW, ADBE" together, and another names "Snowflake, ServiceNow, Atlassian, Workday". These are two expressions of one trade — the AI-disruption-of-enterprise-software debate — and holding both is concentration dressed as diversification. Of the two, ServiceNow has the stronger disclosed forward book ($29.0B of RPO, 98% renewal) and the weaker recent margin trend. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $118.14.

Single biggest risk: the bear case is a claim about 2030 that no 2026 quarter can refute, and the lagging indicators will warn late. At 23.5x the FY2027 consensus, the price already assumes the terminal value survives. Remaining performance obligations of $29.0 billion, current RPO growth of 21% and a renewal rate of 98% are all measures of decisions customers have already made — on three-year contracts, mostly signed before agentic tooling was cheap. The specific mechanism a conviction-80 knowledge-base claim describes — enterprises spinning up bespoke alternatives once agent unit cost falls to roughly 30% of value delivered, and cutting renewals 50-75% — would first become visible in the renewal rate, and by then two years of contracts would already be at risk. ServiceNow's own response has been to spend $8.8 billion buying adjacencies and to borrow $6 billion to do it, which raises the stakes in both directions. There is no number in this file, and none coming on 2026-10-28, that settles it.

Provenance & disclosures

• Traceability: 45 unique tagged knowledge-base claims name ServiceNow or the ticker NOW by entity — 33 matching ServiceNow and 13 matching a case-sensitive standalone NOW, with one overlapping both. This is the largest lane in this batch by a factor of ten against the median. 3 claims sit in the quarantine_misattributed bucket and are excluded outright by policy (two bearish AI-commoditisation claims and one bullish rebuttal). Of the 42 admissible, 12 — 28.6% — come from a channel attributed to Jensen Huang, chief executive of NVIDIA, a commercial partner with which ServiceNow jointly built an AI reasoning model and whose hardware the claims describe as "accelerating" ServiceNow's platform. He is not ServiceNow management, but he is a commercially interested party; all twelve of his claims are bullish and make the same argument. They are EXCLUDED from the conviction pool, quoted at half weight, and disclosed. A CONCENTRATION SENSITIVITY TEST was run and materially changed the reading: removing the twelve takes the lane from 42 to 30 claims splitting roughly 17 bullish, 11 bearish, 2 neutral, weakening net conviction from clearly positive to roughly balanced with a negative recency tilt. Named independent speakers in the pool include Chamath (bearish 80), jordi_visser (bearish 80), Josh Brown (bearish 62, 2026-07-14 — the most recent claim in the lane), Adam Parker (bearish 45, and correctly anticipating the "revenue beat, margin miss" shape that the June quarter then delivered), Gavin Baker (bearish 58), Dan Ives (bullish 82 and 55), Andreas Steno (bullish 45), and jacob_shapiro (bullish 62, disclosing a position); the most operationally specific evidence is an unattributed no_priors cluster dated 2026-04-17 (bullish 80 and 70). A knowledge-base DATA DEFECT is reported: the claim "the 'SaaS apocalypse' is overstated" is tagged stance: bearish, conviction 85 when its content is bullish for ServiceNow — the tagger recorded the stance toward the apocalypse thesis rather than toward the company, and a mechanical stance aggregation would score it backwards. AI laboratory and model names appearing inside three quoted claims are redacted per house rules and the redaction is disclosed rather than performed silently. All quotes are verbatim from the stored claim text apart from those redactions.
• Data as-of: most recent filing-verified financials 2026-06-30, from the 10-Q filed 2026-07-23 (income statement, balance sheet, business-combination note with full purchase-price allocations, debt note with the full coupon schedule, remaining performance obligations, renewal rate, revenue disaggregation and the stock-compensation table) · annual statements fiscal 2025 ended 2025-12-31 from inc_a/bal_a/cf_a, corroborated by the 10-K filed 2026-01-29 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873603 = 2026-08-04T20:00:03Z ($118.14, +3.46%; 50-DMA $106.21; 200-DMA $124.27; RSI 62.6; MACD +2.72) · knowledge-base claims 2026-08-04. Fiscal note: ServiceNow's fiscal year is the calendar year. All per-share figures reflect the 5-for-1 stock split effective 2025-12-17, applied retroactively by both the company and the vendor. All figures come from the Synthos vendor data file for NOW or from the SEC filings in the NOW archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2026-01-29 (fiscal 2025); 10-Q filed 2026-04-23 (March 2026 quarter); 10-Q filed 2026-07-23 (June 2026 quarter); 8-K filed 2026-05-15 (the $4,000,000,000 notes offering, with the full tranche schedule — the single most valuable 8-K in this batch); 8-K filed 2026-05-22 (Item 5.02, officer matters); 8-K filed 2026-07-22 (Item 2.02, second-quarter results — cover document only, Exhibit 99.1 not extracted).
• Where the filings contradicted or corrected the vendor (detailed in Section 8): $8.8 billion of acquisitions (Armis $7.6B on 2026-04-20, Veza $1.2B on 2026-03-02) and $6.0 billion of new debt appear in NO vendor field, because bal_a and cf_a both end 2025-12-31, before either transaction — the June balance sheet is taken from the 10-Q throughout and the vendor's is labelled stale; km_ttm.enterpriseValueTTM rebuilt from $128.090B to $122.950B after establishing that the vendor's $5,950M net-debt figure adds $936M of operating lease liabilities to debt AND omits $4,204M of marketable securities that the 10-Q explicitly counts as liquidity ("totaling $6.7 billion as of June 30, 2026"), restating net debt from $5.950B to $810M and every EV multiple with it; est.ebitdaAvg and est.ebitAvg rejected in full as fixed-ratio fabrications at exactly 35.17% and 31.77% of revenue in every year, against an actual fiscal 2025 EBITDA margin of 22.6%; the FY2023 estimate row discarded as null (epsAvg: 0, netIncomeAvg: 0); returnOnEquityTTM rejected because ratios_ttm reports null while km_ttm reports 0.1377 — the two blocks disagree within the same file, and neither is stable across an $8.8 billion acquisition period; quote.yearHigh/yearLow rejected in favour of tech.hi52/lo52. Where vendor and filing AGREED — worth recording: inc_q June-quarter revenue, operating income, pre-tax income, tax and net income all match the 10-Q income statement to the dollar; the implied share count from market capitalisation (1,033,861 thousand) matches the 10-Q cover figure of 1,033,862 thousand; the 5-for-1 split of 2025-12-17 was applied correctly and consistently by the vendor across all six annual periods — verified against the filing and confirmed clean; seg_prod and seg_geo both sum to reported revenue exactly in every year, with the 62.9% North American home market present; stockBasedCompensationToRevenueTTM reconciles to $1,955M on $13,278M; dividendPerShareTTM of zero is correct.
• Basis note: analyst estimates and earn_cal actuals are non-GAAP; reported diluted EPS is GAAP, and the gap is very large — trailing non-GAAP EPS is $3.754 against trailing GAAP diluted of $1.61, a 2.3x difference driven principally by stock-based compensation of $1,955M in fiscal 2025 (14.7% of revenue) and now also by acquired-intangible amortisation. Every forward multiple in this dive — including the headline 23.5x — is computed on a number that excludes roughly $2 billion a year of real economic cost. The mitigating fact is that buybacks of $2,326M in six months exceeded stock compensation and the share count fell 1.3%. The earn_cal basis is consistent across all five reported rows, so the beat series (+14.7%, +12.0%, +4.0%, +0.0%, +4.7%) is like-for-like and shows beats narrowing sharply — from 14.7% in October 2025 to 0.0% in April 2026 and 4.7% in July. The trailing multiple is shown on both bases (31.5x non-GAAP, 73.4x GAAP diluted).
• Estimate coverage: 31-33 analysts on FY2026 and 26-35 on FY2027 — excellent coverage on the years the valuation uses; 13 on FY2028 EPS, 11 on FY2029 and 5 on FY2030, which is excluded. The FY2028 EPS distribution — low $3.633, mean $6.096, high $9.139, a 2.52:1 low-to-high range — is the widest in this batch and is quoted in Section 6 as the honest measure of how unresolved this name is. The fair-value anchors are deliberately struck on the FY2027 distribution (low $4.392, mean $5.018, high $5.250, 26 analysts) rather than FY2028, because a 2.5:1 range would permit almost any answer.
• Peer note: the vendor peer set contains genuine comparables — Salesforce, Intuit, Palo Alto Networks, AppLovin, Shopify — alongside three semiconductor names (Applied Materials, Lam Research, Qualcomm) and Arista and Uber, which are not. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn. The knowledge base itself repeatedly groups ServiceNow with Salesforce, Workday, Adobe and Snowflake, and Snowflake is in this batch and appears in the same bearish claim — a batch-overlap point made explicitly in Section 11.
• Fair-value caveat: the $80 / $135 / $200 anchors are multiples of the FY2027 consensus non-GAAP EPS distribution — 18x the low of $4.392, 26.9x the mean of $5.018, and 38x the high of $5.250 — cross-checked against the FY2028 distribution. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 24x gives $120, essentially spot; 30x gives $151. The base assumes mild multiple compression, which Section 6b states plainly is a conservatism rather than an analytical claim — we do not believe we can forecast where a multiple that has halved on a philosophical argument settles. The bear case assumes the disruption thesis is right; the bull case assumes it is wrong and a growth multiple is restored.
• Timing: second-quarter results were released 2026-07-22, thirteen days before this dive, and the 10-Q filed 2026-07-23, so the full first half including both 2026 acquisitions is in the data. The next print is 2026-10-28, 85 days away — there is no near-term company-specific catalyst, which is an explicit input to the Watch verdict. 2026-08-04 was a strong day for the stock (+3.46%) inside a broadly higher market; no company-specific news for that date appears in this file.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.