Atlassian Corporation (NASDAQ: TEAM) — The System of Work at a Fire-Sale Price, Where the Cash Flow Belongs to Employees

Independent fundamental research. Report date: 2026-06-13. Fiscal year ends June 30. All figures USD unless noted.

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information, not investment advice. The analysis that follows takes no position and contains no price target — by design.

Verdict: HOLD with a constructive lean / accumulate-on-further-weakness. A genuine 20%-durable-growth franchise — Jira, Confluence and Jira Service Management running the work of ~350,000 organizations at 83% gross margins, 120%+ net retention and >99% retention in its $1M+ cohort — trading at the 2.4th percentile of its own ten-year price-to-sales history (the cheapest a dollar of Atlassian revenue has ever been valued in public markets), ~13.6x forward non-GAAP EPS. But this is not the clean buy that Workday was at a similar de-rate: stock-based comp of 26% of revenue means owner free cash flow (FCF minus SBC) is roughly breakeven, the buyback only mops dilution (the share count still rises), the founders control 85% of the vote on 37% of the economics, have bought zero shares on the open market into a 60% drawdown, and just spent $488M of shareholder cash on a consumer web-browser. Directional fair-value zone ~$110–150 (a re-rate toward ~16–19x forward non-GAAP EPS as GAAP operating profit emerges in FY27 and owner-FCF begins to inflect) vs ~$88 today; genuine value below ~$80; froth above ~$185. Conviction: medium.

Tag: “Cheapest it has ever been — because the owner comes last.”

The setup rhymes with the de-rate that hit Workday, ServiceNow and Salesforce, but it is more violent and lower-quality: TEAM is down ~60% from its ~$222 high into a near-trough own-history valuation, even though the durable engine is accelerating — Cloud revenue crossed $1.1B a quarter and re-accelerated to ~29% YoY in Q3 FY2026, net revenue retention is above 120% and ticking up for several straight quarters, and the $1M+ and $3M+ ARR cohorts are compounding +39% and +54%. The franchise underneath is real: Jira and Confluence are the system of record for an organization’s work, history and process, with switching costs measured in years of embedded workflows and a $4B+ third-party Marketplace ecosystem layered on top. The market is pricing FY2027’s self-inflicted revenue-growth trough (Atlassian is deliberately end-of-lifing its Data Center on-prem product, pulling license revenue forward into FY2026 and guiding a negative Data-Center line in FY2027) and the “AI eats seat-based software” fear as if both were permanent — when the Cloud engine, the migration tail and the rising retention argue they are temporary.

What keeps this a HOLD rather than the BUY I assigned Workday is the quality of the cash and the stewardship of it. Workday’s buyback fully absorbed its SBC and shrank the share count; Atlassian’s offsets only ~57% of a far heavier SBC load, so dilution continues and the headline 27% FCF margin overstates value to outside owners by ~25 points — on an owner basis the company throws off roughly zero distributable cash today. The “cheap” screen is therefore conditional: it only pays off if SBC normalizes from 26% toward the low-teens and headline cash converts to per-share cash (management guides GAAP operating profit to begin in FY2027 — the tell to track). Layer on an unaccountable dual-class structure, an insider tape that is 100% selling (founders $200M+, zero buys) and a $488M browser acquisition outside holders could not veto, and you have a contrarian-value setup with a real franchise and a real governance/quality discount. The framing is value with a quality caveat, not falling-knife and not pristine compounder. Conviction: medium. Flips decisively bullish: two-to-three quarters of Cloud holding ~25%+ with SBC/revenue visibly falling and the diluted share count actually shrinking — proof the owner-cash inflection is real, not promised. Flips bearish: net retention cracking below ~115% or visible seat erosion at large accounts (the AI-disruption thesis showing up in the numbers), or SBC staying at 25%+ while the founders keep deploying cash into adjacencies. At ~13.6x forward earnings a great deal of pessimism is in the price; the asymmetry favors the patient buyer who is honest that the owner is last in line for the cash.

1. Executive Summary

Atlassian is the dominant independent platform for technical and knowledge-team work management — the company behind Jira (issue/project tracking), Confluence (team knowledge/wiki), and Jira Service Management (IT/enterprise service management), plus Bitbucket, Trello, Loom, Compass and the Rovo AI layer. It serves ~350,000 paying customers — 85% of the Fortune 500 are customers, though the F500 is only ~10% of revenue — on an overwhelmingly recurring model at ~83% gross margin. FY2025 (ended June 30, 2025) revenue was $5.22B, up 20%, and the durable Cloud line re-accelerated to ~29% by Q3 FY2026.

The business generates real cash: FY2025 operating cash flow of $1.46B and free cash flow of ~$1.42B (~27% margin) on just $45M of capex — software economics in pure form. The central quality-of-earnings problem is stock-based compensation of $1.36B — 26% of revenue, the highest in the de-rated SaaS cohort — which is why a company with 83% gross margins posts a GAAP operating loss every year since FY2023 (FY2025: −$130M). Net the SBC out and owner free cash flow is roughly breakeven; the $779M FY2025 buyback offsets only ~57% of SBC, so the diluted share count still rises ~1–1.5%/yr. The Rule of 40 reads 52 on headline FCF but ~26 on owner economics — and the gap between those two numbers is the entire investment debate.

The stock is the story. TEAM trades at ~$88, ~60% below its 52-week high of $222.59, at the 2.4th-percentile price-to-sales and 15.6th-percentile composite of its own ~10-year valuation history — i.e., close to the cheapest it has ever been as a public company — on ~13.6x forward non-GAAP EPS and ~4.3x EV/revenue. The de-rate is partly sentiment (the “agentic AI destroys seat-based SaaS” narrative that compressed the whole cohort) and partly self-inflicted: Atlassian’s September 2025 end-of-life of its Data Center product pulls license revenue forward into FY2026 and produces a guided revenue-growth trough in FY2027 (with a negative Data-Center line) before FY2028 re-acceleration. The early operating evidence runs against the bears: Cloud is re-accelerating, retention is rising, ~2/3 of Jira users and 7/10 Confluence users are non-developers (decoupling the TAM from software-engineering headcount), and Rovo AI adoption is inflecting (75% of the F500 have turned it on; credits +20% month-over-month).

This memo evaluates the franchise (a real, financially-visible switching-cost-plus-ecosystem moat that is deepening in the enterprise while eroding among modern dev teams who increasingly choose Linear and Notion), the genuine deceleration (34%→20% and the FY27 trough), the SBC/owner-FCF wedge, the founder-controlled governance (85% of votes on 37% of economics, zero insider buying, a $488M browser acquisition), and the embedded expectations. The analysis below takes no position and sets no price target (see Claude’s Take above for the single, fenced-off exception). The variant question is not business quality — the moat is real — but price × the durability of seat-based work-management economics in an agentic-AI world × whether 26%-of-revenue stock comp ever converts into cash for outside owners.

2. Business Overview

What Atlassian does. Atlassian Corporation (NASDAQ: TEAM) builds collaboration and work-management software for technical and non-technical teams inside organizations. The historical core is two products — Jira (issue/project tracking, born as a developer bug-tracker) and Confluence (a team wiki/knowledge base) — but the company has spent the last several years deliberately repositioning itself from “a bundle of point apps” into a single cloud platform it now markets as the “System of Work,” underpinned by a context layer it calls the Teamwork Graph. Management’s framing is that Atlassian is becoming “an operating system for work” for large enterprises (Investor Day, 2026-05-06). The portfolio is now organized into five collections rather than discrete SKUs: the Teamwork Collection (Jira + Confluence + Loom + Rovo AI bundle — the company’s primary AI-monetization vehicle), the Service Collection (Jira Service Management/ITSM and adjacent employee- and customer-service desks), a Software/Dev grouping (Jira plus Bitbucket, Compass), a Strategy/Planning grouping (Jira Align, Jira Product Discovery, Focus, goals/projects), and newer enterprise/AI add-ons (Guard for security/governance, DX for engineering-productivity measurement, Talent). Loom (acquired 2023, ~$975M) adds async video; Rovo is the AI agent/search/chat layer that runs across all surfaces.

How it makes money — three revenue lines (FY ending Jun 30). Atlassian reports revenue in three buckets:

1. Cloud — recurring SaaS subscriptions, now the growth engine and the majority of revenue. Cloud surpassed $1.1B in Q3-FY26 (Mar-2026 quarter) and re-accelerated to ~29% YoY (Q3-FY26 call, 2026-04-30). On a full-year basis Cloud is roughly ~63% of revenue (FY25 10-K).
2. Data Center — self-managed, on-premise/private-cloud term licenses (the successor to the discontinued Server product, which Atlassian end-of-lifed in February 2024). Data Center is recognized partly upfront under ASC 606 and is roughly ~one-third of revenue. Crucially, Data Center is now itself being wound down: in September 2025 Atlassian announced “Ascend,” the end-of-life of Data Center by March 2029 (Investor Day, 2026-05-06), forcing the remaining on-prem base toward Cloud.
3. Marketplace & Other — Atlassian’s ~10–15% cut of third-party app sales on the Atlassian Marketplace plus professional/advisory services; a small (~4–5%) but high-margin, ecosystem-reinforcing line.

Total revenue: FY21 $2.09B → FY22 $2.80B (+34%) → FY23 $3.53B (+26%) → FY24 $4.36B (+23%) → FY25 $5.22B (+20%); FY26 9-month revenue $4.81B, with Q3-FY26 total revenue $1.787B, +32% YoY (EDGAR XBRL, RevenueFromContractWithCustomerExcludingAssessedTax; Q3-FY26 10-Q filed 2026-05-01). Gross margin runs ~82–83% GAAP / ~88% non-GAAP (FY26 guidance, Investor Day). Revenue is overwhelmingly recurring: Cloud is subscription; Data Center is term-license (annual, increasingly 1-year as EOL approaches); only Marketplace/services is partly transactional. Net revenue retention is 120%+ and has ticked up for several consecutive quarters (Q3-FY26 call) — the single most important quality signal in the model.

Customers and go-to-market. Atlassian has ~350,000 paying customers spanning essentially every industry, and 85% of the Fortune 500 are customers — but the F500 is only ~10% of revenue (Investor Day), which is both a runway statement and a tell that the base is still mid-market-weighted. The classic motion is product-led, “low-touch” / bottoms-up: a generous free tier and self-serve sign-up funnel let individual developers and teams adopt Jira/Confluence with no salesperson, then land-and-expand as usage spreads across the org. This produced billions in revenue and 350k customers with famously low customer-acquisition cost. That model is now deliberately changing. Under CRO Brian Duffy (hired ~2024), Atlassian has bolted a traditional enterprise sales force onto the PLG base — quota carriers grew from 117 (FY22) to ~400 (FY26) — plus first-ever GSI partnerships (Accenture, Deloitte, PwC) to reach the C-suite (Investor Day). This is the central operational transition: the funnel still feeds the bottom of the pyramid for free, while a real enterprise motion monetizes the top. The proof is in the upmarket cohorts: $1M+ ARR customers passed ~600 (+39% YoY, 6x in four years), >99% retention; the $3M+ cohort grew ~54% YoY (10x in four years); deals over $3M surged ~79% YoY in Q3, deals over $5M +54% YoY, with ASPs up ~22% (Investor Day).

Geographic mix. Roughly ~45% of revenue is United States, with the balance international (Americas-ex-US, EMEA, APAC) per the FY25 10-K disaggregation; management is explicitly investing in under-penetrated geographies (UK, France, India). Founders Mike Cannon-Brookes and Scott Farquhar built the company in Sydney; it reincorporated in Delaware in 2022. Cannon-Brookes is sole CEO of record on the FY26 calls (Farquhar stepped back from the co-CEO role in 2024 to a board/special-advisor capacity — OPEN QUESTION: confirm exact current title in proxy). New CFO James Chuong joined ~April 2026.

Verdict (Business Overview). A high-quality, overwhelmingly-recurring, ~82–83%-gross-margin SaaS platform with rare durability — 20%+ revenue growth at >$5B scale, 120%+ NRR, and a 350k-customer install base it is steadily converting from cheap free/standard seats into expensive enterprise commitments. The model is mid-transition on two fronts simultaneously: (a) a forced Server→Data Center→Cloud migration that is now in its final, lumpy Data-Center-EOL leg (creating the FY27 revenue-recognition “trough” management flagged), and (b) a PLG→PLG-plus-enterprise-sales go-to-market rebuild. Both transitions are progressing on the evidence, but they make the reported numbers noisier than the underlying business, and the second one structurally raises the cost-to-serve that the bottoms-up model was prized for avoiding.

3. Industry Dynamics

Market structure. Atlassian sits at the intersection of several large, adjacent software categories: agile project/work management (Jira’s core), knowledge management/collaboration (Confluence), IT & enterprise service management / ITSM (Jira Service Management vs. ServiceNow), developer tooling/source control (Bitbucket vs. GitHub/GitLab), and increasingly AI-orchestration of work (Rovo). None of these is a tidy oligopoly; each is contested by a different cast. The result is a structurally fragmented, intensely competitive software landscape where the most dangerous competitor — Microsoft — bundles overlapping functionality (Teams, Planner, Loop, Azure DevOps, GitHub Issues/Projects) into a suite many enterprises already pay for. Switching costs and ecosystem lock-in exist (see §7.3) but are product-and-deployment-specific, not industry-wide moats that protect all incumbents.

Market size / TAM. Management materially re-cut its serviceable addressable market to ~$140B at the May-2026 Investor Day, up from the ~$67B figure cited in prior years — the upsizing reflects the move beyond developers into the full knowledge-worker base (Atlassian frames the prize against ~1 billion global knowledge workers, of whom ~900M are non-technical) and the new AI/agent and enterprise-service categories. INTERPRETATION: a doubling of stated SAM in one year is a narrative as much as a measurement — treat the absolute number skeptically (§0 rule 8). The directionally-correct point stands: Atlassian’s reach into finance/HR/legal/marketing/ops “business teams” is genuinely larger than its developer heritage, and Confluence (7 of 10 users non-developer), JSM (>3/4 non-developer) and even Jira (~2/3 non-developer) confirm the base is already majority non-technical.

Secular tailwinds. (1) Digital transformation / “every company is a software company” — technology workflows proliferating into non-tech functions expands the canvas for a “system of work.” (2) Growth in technical and knowledge-worker headcount historically drove seat growth — the bull case is that the world keeps adding the workers Atlassian charges per-seat for. (3) Cloud migration — the secular shift off on-prem is a tailwind Atlassian is forcing via Server and Data Center EOL, converting low-monetization legacy seats into higher-ARPU Cloud subscriptions (DC→Cloud migrants land in premium/enterprise tiers 93% of the time and grow ~1.5–2x over the three years post-migration, per Investor Day).

The AI disruption question (the central structural debate). The bear thesis is that AI is an industry-level threat to Atlassian’s seat-based model: if AI coding agents write software, enterprises need fewer developer “seats,” and seat-based pricing deflates; AI-native startups could rebuild work-management tools faster and cheaper. This is the same disruption fear hanging over every seat-priced application-software vendor. Management’s counter-framing (Investor Day; Q3 call) is that AI is “one of the best things that ever happened to Atlassian” because: (a) seat growth excluding migrations is still compounding with no compression visible; (b) the bottleneck in an agentic world shifts from writing work to orchestrating and contextualizing it, which is exactly what the platform + Teamwork Graph provides (“context is the only anchor to avoid chaos”); and © AI creates new monetization — consumption-based Rovo credits (growing 20% month-over-month), agent runs, and net-new C-suite products (DX, Talent, Focus). INTERPRETATION: both can be true, and the timing is the variable. Near-term, the data (Q3 seat expansion in stand-alone Jira, NRR ticking up, no compression) supports management. Medium-term, the seat-based-software business model genuinely faces a re-rating risk if value migrates from “number of human seats” to “amount of agent compute,” and Atlassian’s pivot to usage/consumption meters is partly an admission of that. The honest read: AI is simultaneously a real risk to the pricing unit and a real expansion of the workflow-orchestration TAM — it is not obviously net-negative for TEAM, but it is the dominant source of uncertainty in the industry.

Capital cycle (Marathon lens). Application/collaboration SaaS is in a mature-to-late capital-cycle phase: a decade of cheap capital and a 2020–21 funding boom flooded the category with VC-backed challengers (Monday, Asana, ClickUp, Notion, Linear, Airtable, Smartsheet, Coda, Height, Shortcut), fragmenting demand and competing on UX. The 2022–24 rate reset has thinned the herd and shifted survivors toward profitability — a modestly favorable supply-side development for incumbents. But the AI wave is precisely the kind of technology disruption that breaks the normal capital cycle (per the framework): it lowers the cost to build a competing tool and is attracting a fresh wave of AI-native capital, re-fragmenting the supply side just as it was consolidating. Net: the supply side is not cleanly favorable.

Regulation. Light. Data-residency, privacy (GDPR), security/compliance (FedRAMP — Atlassian is in Google Cloud / FedRAMP Moderate), and enterprise data-governance requirements are barriers to entry that favor scaled incumbents (a tailwind for Atlassian vs. startups) more than they are a regulatory threat. No reimbursement, rate, or licensing regime governs the category.

Verdict (Industry Dynamics). A structurally mixed-to-good industry, not a great one. Positives: large and genuinely expanding TAM, durable digital-transformation demand, high gross margins, recurring revenue, and compliance barriers that favor scaled players. Negatives: chronic fragmentation, a single overwhelming bundled competitor (Microsoft) able to give adjacent functionality away, low barriers to building a competing tool, and an AI transition that simultaneously expands the opportunity and threatens the seat-based pricing unit while re-energizing the challenger supply side. This is an industry where firm-specific competitive advantage — not industry structure — has to carry the thesis. On that test the industry is a backdrop, not a moat.

4. Competitive Position

The question. Atlassian compounds 20%+ at >$5B with 120%+ NRR and >99% retention in its largest cohorts — outcomes that, per the Greenwald tests (sustained high retention, share stability at the top, pricing power), imply real competitive advantage. The task is to name the mechanism, pressure-test it, and judge durability against a crowded field.

Candidate moats, pressure-tested:

(a) Switching costs — the primary, real moat (demand-side customer captivity). This is the strongest and best-evidenced advantage. Jira and Confluence become the system of record for an organization’s work, history, and process — years (often 15–22, per the two anonymized Investor-Day case studies and the Cisco/Canal+ panel) of issues, workflows, automations, permissions, integrations and institutional knowledge accrete inside them. Ripping them out means migrating data, retraining thousands of users, rebuilding hundreds of custom workflows/service desks, and re-integrating the surrounding tool stack — high cost, high error-risk, and high political risk. The Canal+ CTO’s answer to “what if Atlassian went away?” — “we’d be back to Excel files and hours of meetings” — is exactly the captivity Greenwald describes. The financial proof that this is a real moat (not a narrative): >99% retention in the $1M+ cohort, 120%+ NRR at scale, and DC→Cloud migrants growing 1.5–2x post-migration. A moat is only a moat if economics would deteriorate without it; here, churn economics would visibly collapse without the embedded-workflow lock-in. Verdict on (a): genuine, durable, deepening in the enterprise.

(b) Marketplace ecosystem + Forge platform — network effects / scale-plus-captivity, the second real moat. The Atlassian Marketplace has surpassed $4B in cumulative app sales (March 2025, accelerating), with ~5,000+ apps from ~1,250+ partners (Atlassian developer blog 2025–26; appmarketplace.com). This is a genuine two-sided network effect: more customers attract more developers, whose apps make Jira/Confluence stickier and harder to replace, which retains more customers. Enterprises frequently run business-critical apps purchased through the Marketplace, deepening switching costs beyond Atlassian’s own code. The Forge developer platform (cloud app-hosting; 0% revenue share up to $1M lifetime through 2026 to seed supply) is Atlassian’s bid to keep that ecosystem captive as it moves to Cloud. Pressure-test: this is a real but bounded moat — it reinforces (a) rather than standing alone, and a determined competitor (GitHub, ServiceNow) has its own marketplace. But at $4B cumulative and accelerating, it is a meaningful, hard-to-replicate asset. Verdict on (b): real network-effect moat, reinforcing the switching-cost moat.

© Bottoms-up viral distribution / low CAC — historically a cost-advantage moat, now eroding by management’s own choice. The free-tier, self-serve, no-salesforce motion gave Atlassian structurally lower customer-acquisition cost than enterprise-sales-led rivals — a durable cost advantage that funded R&D over S&M. But this advantage is being deliberately diluted: quota carriers 117→400, new GSI partnerships, “slowing down deal cycles” to land bigger contracts (Investor Day). The PLG funnel still operates (and remains a genuine edge versus pure top-down vendors), but the cost-to-serve is rising as Atlassian climbs into the enterprise. Verdict on ©: a real and differentiated advantage, but narrowing — and partly self-inflicted as the company chases enterprise dollars.

(d) Brand / intangibles with developers — eroding at the top. Atlassian’s developer brand was once an asset; today it is a liability among modern teams. Jira is widely criticized by developers as bloated, slow, and over-configured (“a single Jira ticket might have 30 fields”; “Time to Interactive 1.5–3 seconds”), and a cohort of fast-growing, design-led challengers — Linear (keyboard-first, “the new Jira,” loved by startups and modern eng teams) and Notion (flexible docs/wiki eating into Confluence) — are taking mindshare at the leading edge (multiple 2025–26 third-party comparisons; Runtime, “How Linear became the new Jira”). Verdict on (d): not a moat; a vulnerability at the top of the market.

Direct competitor read:

• Microsoft — the existential bundled threat: Azure DevOps + GitHub (Issues/Projects/Copilot) attack dev tooling; Teams/Planner/Loop attack collaboration; all bundled into M365/E5 the customer already owns. Microsoft can give adjacent functionality away. Atlassian’s defense is depth and the integrated system-of-work + Teamwork Graph, not price.
• GitLab / GitHub — single-vendor DevSecOps pipelines competing for the developer workflow; GitHub’s Copilot/agent push is the AI-native edge.
• ServiceNow — the dangerous mirror image: as Atlassian pushes JSM/Service Collection up into enterprise ITSM, ServiceNow pushes down. Atlassian’s claimed “largest-ever quarter for competitive displacements from a major ITSM provider” (Q3-FY26 call) is the live battleground — Atlassian winning on value, modern UX, and AI vs. ServiceNow’s incumbency and breadth. Service Collection passed $1B ARR, +30% YoY, 75% of F500 use it, 60% use it outside IT.
• Monday.com / Asana / ClickUp / Smartsheet — mid-market work-management; competing on UX and breadth, mostly below Atlassian’s enterprise sweet spot.
• Notion / Linear / Coda / Airtable — the modern-team challengers taking the leading-edge mindshare Atlassian’s brand once owned (the §(d) erosion).
• Zendesk / Freshworks — customer-service/ITSM challengers Atlassian now meets as it adds customer service management.

The synthesis — a moat eroding at the top and deepening at the bottom. The honest, two-sided read: Atlassian is losing developer love at the leading edge (Linear/Notion among startups and modern teams) while its enterprise lock-in deepens (>99% retention, 120%+ NRR, 600+ $1M customers, ITSM displacements, GSI-led C-suite penetration). Because revenue concentrates in the large, sticky, embedded accounts — not the cohort most tempted by Linear — the financial moat is, for now, widening even as the brand moat at the frontier narrows. The risk is temporal: today’s startups on Linear are tomorrow’s enterprises that never adopt Jira, slowly starving the bottom of the funnel. The AI/Teamwork-Graph bet is management’s attempt to convert the moat from “system of record” to “system of intelligence” — owning the context layer that agents (Atlassian’s and third parties’ via MCP) must query. INTERPRETATION: if real, this is the strongest version of the moat — a data/context advantage that compounds with usage (Marathon’s “embedded in the workflow” pricing power). But it is early, partly narrative, and unproven at the unit-economic level; treat it as a promising hypothesis, not an established advantage (§0 rule 8).

Verdict (Competitive Position). Atlassian has a genuine, durable moat — primarily demand-side switching costs (embedded workflows, data, and process as the system of record), reinforced by a real Marketplace network effect ($4B+ cumulative app sales) and a still-differentiated low-CAC PLG funnel. In Greenwald’s taxonomy this is customer captivity reinforced by ecosystem scale — the more robust end of the spectrum. The moat is deepening where the money is (enterprise) and eroding where the future is (modern dev teams choosing Linear/Notion), and its long-term durability now hinges on whether the AI/Teamwork-Graph context layer genuinely re-anchors Atlassian in an agentic world or whether seat-based work-management proves a category AI commoditizes. Durable today; the verdict on 2030 is open and rests on the AI pivot.

5. Growth History and Forward Opportunities

The headline number is a lie of timing. Atlassian reported Q3-FY26 (qtr ended Mar-2026) total revenue of $1.787B, +32% YoY — a re-acceleration that, taken at face value, would suggest a 20%-grower has found a second gear. It has not. ~$50M of the Q3 beat was upfront term-license revenue pulled forward from future periods because of the Data Center end-of-life decision (Q3-FY26 call, 2026-04-30: “we recognized approximately $50 million more in upfront term license revenue”). Management has explicitly told investors not to anchor on this: at the May-2026 Investor Day they retired their own prior “3-year 20%+ CAGR through FY27” target, stating plainly: “Number one, we’re expecting a trough in total revenue growth in FY '27. Number two, we expect that to significantly reaccelerate in FY '28 as we lap that data center effect… we now expect negative data center revenue growth in '27” (Investor Day, 2026-05-06). FY26’s reported ~30%+ growth is borrowing from FY27, which will print a deliberate trough.

Revenue history (FY ends Jun 30) — FACT:

The underlying secular trend is decelerating: 34% → 26% → 23% → 20%. That is the natural law of large numbers compounded by a $5B base. The FY26 optical re-acceleration does not break the trend; it defers part of it into a negative FY27 DC line.

Growth by line (the composition is the whole story):

• Cloud (the durable engine): surpassed $1.1B/quarter in Q3-FY26 and re-accelerated to ~29% YoY (Q3-FY26 call). This is the line that matters. It is genuinely re-accelerating — driven by seat expansion in core Jira (management called out “strong seat expansion in Jira” as the key driver), the migration of large DC estates into premium/enterprise cloud tiers (93% of migrating DC customers land in premium/enterprise — Investor Day), and AI cross-sell. INTERPRETATION: if you strip the accounting noise, Cloud at ~29% on a >$4B annualized base is the genuinely impressive datapoint and the only one a long-term owner should weight.
• Data Center (the melting line): Q3-FY26 DC grew ~+44%, but this is terminal. EOL “Ascend” was delivered Sept-2025 (final EOL March 2029). Because customers cannot buy a 3-year DC term deal a year before a 3-year EOL window, term revenue front-loads into FY26 and goes negative in FY27. DC retention is “incredibly robust” and is converting to mid-to-high-single-digit cloud growth contribution, but the DC line itself is being euthanized on purpose.
• Marketplace & other: high-margin app/plugin ecosystem revenue, smaller and steadier; not a swing factor.

Customer / seat / retention metrics (FACT):

• ~350,000 customers; >600 customers at $1M+ ARR, +39% YoY, >99% retention — the up-market motion is real and the high end is sticky.
• $3M+ ARR cohort grew 10x in 4 years, +54% YoY in the last year alone (Investor Day) — the enterprise land-and-expand engine.
• NDR/NRR maintained >120% and ticked up for the third-to-fourth consecutive quarter (Q3-FY26) — expansion is accelerating, not fading, which directly contradicts the seat-compression bear case so far.
• Seat growth excluding migrations is still positive — management spent the Investor Day directly rebutting “seats are going to disappear,” showing cloud seat growth ex-migrations is intact. OPEN QUESTION: this is the single most important forward debate (see Risk matrix and Variant Perception).

Forward drivers (ranked by credibility):

1. Cloud migration tail (high credibility): “the migration itself still has a ways to go in terms of the number of seats… in the data center.” Migrating customers grow ARR 1.5–2x over the 3 years post-migration, landing 93% in premium/enterprise. This is a multi-year, visible, mechanical tailwind to Cloud — and it is exactly what FY28’s reacceleration relies on.
2. Enterprise up-market motion (high): CRO Brian Duffy’s enterprise sales build (quota carriers expanded from ~117 to ~400) is driving larger deals; ASP +22%, longer commitment durations, $3M cohort +54%.
3. AI / Rovo cross-sell (medium, early but inflecting): Rovo AI-credit usage +20% month-over-month; 75% of the Fortune 500 have turned Rovo on; Rovo customers grow ARR at ~2x the rate of non-Rovo customers; Teamwork Collection passed 1M seats / 1,000 customers in <6 months. Monetization is still mostly via the Teamwork Collection bundle (10x more credits on upgrade) plus nascent consumption meters. INTERPRETATION: the leading indicators are strong but ARR contribution is still small; this is a “show me FY27-28” driver.
4. Non-developer expansion (medium-high): 7 of 10 Confluence users and ~2/3 of Jira users are non-developers — decoupling the TAM from software-engineering headcount, the most important structural rebuttal to the “AI shrinks dev seats” bear thesis.
5. International + SAM re-cut: SAM raised to ~$140B (from ~$67B) at Investor Day — a self-serving but defensible re-rack given the platform/AI surface expansion; ~6M target companies are not yet customers.

VERDICT — high-quality growth, temporarily wearing an ugly accounting costume. The durable engine (Cloud +29%, NDR >120% and rising, $1M+/$3M+ cohorts compounding 39-54%, >99% gross retention) is high-quality: organic, software-margin, expansion-led, increasingly enterprise and increasingly non-developer. The blemishes are real — secular deceleration is the long-run reality (34%→20%), the FY26 print is inflated and FY27 will trough with a negative DC line, and AI monetization is still a leading-indicator story not a P&L story. But the quality of the dollar of revenue is high and getting higher (premium/enterprise mix shift). This is high-quality growth that the market is being asked to underwrite through a one-year, self-inflicted accounting trough — the quality is in the Cloud line; the risk is whether investors hold through FY27.

6. Financial Quality

Verdict up front: High-growth, asset-light, gross-margin-rich — and GAAP-unprofitable by design, because stock-based compensation (~26% of revenue) consumes the entire operating profit and most of the free cash flow. The economics do scale on a cash basis, but the “non-GAAP profitability” the Street underwrites is largely SBC add-back; the honest owner-earnings number is near breakeven. A sound business with a real cash engine, reported through the most generous lens in its cohort.

The growth and margin structure is genuinely good — on the surface

Revenue compounds at a rate few software companies of this scale match: $2,089M (FY21) → $2,803M (FY22) → $3,535M (FY23) → $4,359M (FY24) → $5,215M (FY25), a ~26% five-year CAGR, with FY26 nine-month revenue of $4,806M up 25% YoY and Q3-FY26 up 32% to $1,787M (10-Q, Mar-31-2026). Gross margin is best-in-class: FY25 gross profit $4,320M on $5,215M = 82.8% (FY25 10-K / ARS), and 84.8% on a trailing basis (public market data). This is a textbook scale-economics-plus-low-marginal-cost SaaS profile: each incremental dollar of subscription revenue drops ~83 cents of gross profit, and the bottoms-up, self-serve distribution model keeps sales & marketing intensity below peers.

…but the company has run a GAAP operating LOSS every year since FY23

GAAP operating income: +$141M (FY21), +$70M (FY22), then −$345M (FY23), −$117M (FY24), −$130M (FY25) — three straight years of operating losses on a growing and 82%-gross-margin revenue base (XBRL, reconciled to 10-K). GAAP net loss every year of the window: −$579M, −$520M, −$487M, −$301M, −$257M (FY21–FY25), and −$192.9M for 9mo-FY26 (10-Q). A company cannot post operating losses on 83% gross margins unless operating expense is the problem — and the single largest, fastest-growing operating expense is stock-based compensation.

Stock-based compensation is the central quality-of-earnings issue

SBC: $341M (FY21) → $525M (FY22) → $948M (FY23) → $1,081M (FY24) → $1,362M (FY25) = 26.1% of revenue (cash-flow statement, 10-K). For 9mo-FY26 SBC was $1,212M, run-rating ~$1.6B/yr and still ~25% of revenue. This is the highest SBC/revenue ratio in the large-cap collaboration-software cohort — versus roughly 21% at ServiceNow-adjacent peers and ~17% at Workday on comparable measures. SBC is a real economic cost — it transfers ownership from existing holders to employees — but it is non-cash, so it inflates operating cash flow while it depresses (correctly) GAAP earnings.

The non-GAAP bridge is almost entirely SBC. Management guides to and is measured on a non-GAAP operating margin (FY27 commitment “25%+”, per Q2-FY26 call, Feb-5-2026), versus the −2.5% GAAP operating margin in FY25. The ~27-point bridge from GAAP to non-GAAP operating margin is built primarily from (a) adding back all SBC (~26 pts) and (b) amortization of acquired intangibles (D&A was $101M for 9mo-FY26, rising with the BCNY/DX deals). In other words, the “profitability” investors pay for is the profitability that exists only if you pretend equity issued to employees is free. It is not free — see owner-FCF below.

Free cash flow is real and large — but “owner free cash flow” is near zero

Operating cash flow has scaled well: $790M (FY21) → $821M → $868M → $1,448M (FY24) → $1,460M (FY25). The business is genuinely asset-light: FY25 capex was only $45M (~0.9% of revenue), so FY25 FCF ≈ $1,415M (~27% FCF margin). That is a strong headline.

But subtract the SBC that the FCF figure adds back and never charges:

• FY23: FCF ≈ $843M − SBC $948M = −$105M owner-FCF
• FY24: FCF ≈ $1,425M − SBC $1,081M = +$344M
• FY25: FCF ≈ $1,415M − SBC $1,362M = +$53M owner-FCF

On an owner basis — treating SBC as the cash-equivalent cost it economically is — Atlassian generated roughly breakeven free cash flow in FY25. The buyback (FY25 $779M, below) offsets only ~57% of the year’s SBC, so the share count still rose. This is the crux: the cash FCF is real, but a quarter of it is being recycled to employees as equity, and the headline FCF margin overstates value creation to outside shareholders by ~25 points. (FACT on the numbers; INTERPRETATION on “owner-FCF” framing.)

FCF quality: not flattered by billings timing — if anything, the opposite right now

A common SaaS trick is OCF flattered by deferred-revenue inflows. Atlassian is currently showing the reverse: in 9mo-FY26, the deferred-revenue change was a −$96.8M use of cash (vs. +$253.5M source in 9mo-FY25), a ~$350M adverse swing that hurt reported OCF (10-Q cash flow). This is a rev-rec timing artifact of the Data Center end-of-life, not deteriorating demand — gross deferred-revenue additions (billings proxy) still grew +16% (to $4,736M for 9mo-FY26 from $4,084M), and RPO grew +37% YoY to ~$4.0B (Q3-FY26 call / 10-Q). So the billings engine is healthy; the OCF line is being understated by the migration, not overstated. 9mo-FY26 OCF of $874M is down from $1,085M YoY — entirely explained by the deferred-revenue swing and higher cash taxes/working capital, not a collapse in cash generation. (FACT.)

Net revenue retention and Rule of 40

• Net revenue retention: Atlassian no longer prints a single clean dollar-based NRR each quarter; management increasingly directs investors to ARR and cohort metrics. Disclosed color (Investor Day, May-6-2026; Q3 call): $1M+ ARR customers grew +39% YoY with >99% gross retention, the $3M+ cohort +54%, and Rovo (AI) customers expanding ARR at ~2x non-Rovo. (FACT on the cohort figures; OPEN QUESTION on a single consolidated NRR — the shift away from a headline NRR is itself a disclosure-quality flag, INTERPRETATION.)
• Rule of 40: revenue growth ~25% (9mo-FY26) + FCF margin ~27% (FY25) ≈ 52 on the headline FCF basis — comfortably above 40. But on an owner basis (FCF-minus-SBC margin ≈ +1% in FY25), Rule of 40 ≈ 26 — below the threshold. Which number is right depends entirely on how you treat SBC; the gap between “52” and “26” is the entire thesis. (INTERPRETATION.)

ROE/ROIC are not meaningful — use the right lens

GAAP is loss-making and equity is thin ($879M at Mar-31-2026, down from $1,346M at Jun-30-2025 as buybacks retire equity faster than retained losses accumulate), so ROE (public market data −19%) and ROIC are mechanically negative/meaningless and should be disregarded. The correct lenses for a self-funding, asset-light SaaS compounder are: FCF margin (~27% headline / ~1% owner), Rule of 40 (52 / 26), gross margin (83%), and dilution net of buyback (+~1–1.5%/yr). On those, the verdict is: excellent gross economics and cash conversion, undermined by an SBC load that captures most of the value for employees rather than owners. (INTERPRETATION.)

Balance sheet and liquidity — meaningfully thinner after the FY26 spending

• Cash & equivalents: $2,513M (Jun-25) → $2,322M (Sep-25) → $1,158M (Dec-25) → $1,136M (Mar-26). The ~$1.16B Sep→Dec drop is resolved (below).
• Marketable securities went to ZERO: $424.3M at Jun-30-2025 → $0 at Mar-31-2026 — the company liquidated its entire marketable-securities book (9mo-FY26 “proceeds from sales of marketable securities” $352M plus maturities $144M) to help fund acquisitions and buybacks. Total liquidity is therefore now essentially just the $1,136M of cash (vs. ~$2.9B cash+securities a year earlier) — a material de-liquefication. (FACT, 10-Q balance sheet.)
• Debt: $1.0B of senior notes — $500M @ 5.250% due May-15-2029 + $500M @ 5.500% due May-15-2034, both issued May-15-2024 (carrying value $989M at Mar-26). Note: this was a FY2024 financing, not a new Sep-2025 offering. Leverage is modest (~0.7x cash FCF), but net cash has shrunk from ~$1.9B to ~$0.1B in three quarters.
• Equity: $879M (Mar-26). Thin, and shrinking by design via buybacks.

§7.5 Verdict: The economics scale on a cash basis — 83% gross margin, ~27% headline FCF margin, ~0.9% capex, 25%+ growth — and that is a genuinely good business. But the quality of earnings is the lowest-grade variety in its cohort: GAAP losses every year since FY23, “profitability” that exists only after adding back the highest SBC/revenue ratio (~26%) in the peer set, and an owner-FCF that nets to roughly breakeven once that equity dilution is honestly charged. The balance sheet, formerly a fortress, has been drained to ~$1.1B cash with marketable securities at zero. Good business, generously reported; the cash is real, but most of the reported value accrues to employees, not owners.

7. Capital Allocation

Verdict up front: A serial bolt-on acquirer that has stepped up both M&A and buybacks in FY26 — but the buyback is overwhelmingly a dilution-mop (it offsets only ~57% of SBC and the share count still rises), and the FY26 acquisitions (BCNY/“Arc browser” $488M, DX $720M) are unproven, adjacency-stretching bets funded by draining the balance sheet. Capital allocation is active, not obviously value-accretive. No dividend, by design.

Free cash flow generation and philosophy

The business throws off ~$1.4B of headline FCF (FY25) and almost no capex, so management has real discretion. Its stated philosophy is reinvest-in-R&D-first, repurchase-to-offset-dilution, and acquire bolt-on technology/talent; no dividend (FACT — none declared; payout ratio 0). The problem is that the largest “use” of cash — SBC — is buried in the operating line, and the second-largest — buybacks — largely just neutralizes it. Net distributable return to shareholders is therefore much smaller than the headline FCF implies.

M&A history — frequent, bolt-on, accelerating, and increasingly adjacency-stretching

Atlassian’s M&A is a steady cadence of tuck-ins, with two notably larger and stranger deals in FY26 (all FACT; prices per 10-Q / 8-K / company releases):

The pattern: most deals are sub-$500M tuck-ins that fold into existing products. But Loom (~$975M, 2023), BCNY (~$488M, 2025) and DX (~$720M, 2025) together are ~$2.2B of larger bets in three years, and BCNY in particular — buying a consumer/enterprise web browser business — is a meaningful departure from the Jira/Confluence/ITSM core. These produced $2,303M of goodwill at Mar-31-2026 (up from $1,304M a year earlier — i.e., ~$1.0B of new goodwill in nine months) and rising intangible amortization (10-Q). No deal has yet demonstrated a clear return; the larger ones (Loom, BCNY) are show-me. (FACT on prices/goodwill; INTERPRETATION on “unproven/adjacency-stretch.”)

Resolving the Q2-FY26 (Sep→Dec) cash drop — not a buyback spike, it was M&A + buyback together

The ~$1.16B cash decline from Sep-25 ($2,322M) to Dec-25 ($1,158M) is fully explained:

• Acquisitions closed in the quarter: BCNY $481.5M cash (Oct-20) + DX $720.4M cash (Nov-10) = ~$1.20B of acquisition cash in Q2-FY26 alone (10-Q: “Business combinations, net of cash acquired” $1,228,875K for 9mo, essentially all in Q2).
• Buybacks continued (~$0.5B in Q2, part of the 9mo $1.5B).
• Partly funded by liquidating the $424M marketable-securities book and drawing down cash.

So the dramatic Q2 cash fall is M&A-driven, not a one-off buyback surge — but the combination of ~$1.2B M&A + ~$1.5B buyback + zero securities sales over nine months is what took total liquidity from ~$2.9B to ~$1.1B. (FACT.)

Buybacks — real dollars, but mechanically a dilution-offset, not net capital return

Repurchases (PaymentsForRepurchaseOfCommonStock): $0 (FY21), $0 (FY22), $150M (FY23), $395M (FY24), $779M (FY25), and ~$1.5B in 9mo-FY26 (≈$1.0B in Q3 alone) at an average price of $100.51/share (9mo) and $85.04 in Q3 (10-Q). All open-market. A new 2025 Share Repurchase Program has ~$2.2B remaining authorization at Mar-31-2026.

The decisive test of whether this is capital return or capital recycling: the diluted share count still RISES. Weighted diluted (=basic, given GAAP losses) shares went from 249.7M (FY21) to 261.8M (FY25) — net dilution of ~1–1.5%/yr despite the buybacks. In FY25, $779M of repurchases offset only ~57% of the year’s $1,362M of SBC. The buyback is therefore not returning capital to existing owners; it is partially mopping up the dilution from paying employees in stock, and not even fully. The FY26 acceleration ($1.5B in 9 months) finally outpaces SBC enough that the count may flatten — but it is being funded by draining the balance sheet and selling the securities book, while buying at $100.51 average against a current ~$88.52 price (i.e., recent repurchases are modestly underwater). (FACT on share count and amounts; INTERPRETATION on “dilution-mop.”)

R&D and S&M intensity

• R&D is extremely high — historically ~45–50% of revenue, among the highest of any scaled software company; this is where the cash and a large share of the SBC go, and it is the bull case (the product velocity — Rovo/AI, Teamwork Graph, the Collections bundle — is real). But at ~50% of revenue it is also why GAAP operating income is negative.
• S&M is structurally low for the revenue scale — the bottoms-up, land-and-expand, self-serve model means Atlassian spends far less on sales than ServiceNow/Workday-style enterprise peers. This is a genuine differentiator and a real cost advantage in distribution (a Greenwald demand-side captivity/low-CAC edge). (FACT on the model; the exact FY25 opex split should be reconciled to the 10-K income statement by the Lead.)

Governance overhang on capital allocation

Two co-founders control ~85% of the vote (below) through 10-vote Class B super-voting shares while owning a minority of the economics. Outside Class A holders have no mechanism to discipline capital allocation, M&A, or compensation. Every large bet (Loom, BCNY, DX, the SBC budget) is effectively a founder decision. (FACT / INTERPRETATION.)

§7.6 Verdict: Management is an active allocator with a real R&D edge and a genuinely cheap distribution model — but the FY26 record is not obviously value-accretive. The buyback, though large and accelerating, is fundamentally a dilution-offset (share count still rising; recent purchases underwater at a $100.51 average); the two big FY26 deals (BCNY/browser $488M, DX $720M) are unproven, adjacency-stretching, and were funded by draining liquidity from ~$2.9B to ~$1.1B and selling the entire securities book. Combined with founder voting control that removes any external check, capital allocation is best described as competent-but-unproven, and tilted toward employees and empire over outside owners. Has management allocated capital intelligently? On R&D and the distribution model, yes; on the FY26 M&A-plus-buyback-plus-balance-sheet-drain, not yet demonstrated — show-me.

7.1 Insider / Form 4 Read (SEC Filings Sweep)

Binary verdict: NO open-market purchases. Zero. The entire insider tape over the trailing ~13 months (May-2025 → Jun-2026, 165 Form 4s) is selling, 100% under 10b5-1 plans, overwhelmingly by the two co-founders.

• Code-P (open-market buy) count: 0 across all 165 Form 4s. There is no insider conviction-buying signal whatsoever. (FACT.)
• Sell tape: ~972 sale (code S) transaction lines, plus routine option-exercise/conversion © and small grant (A) entries. Every sale-bearing filing references a Rule 10b5-1 trading plan (120/120 of Cannon-Brookes’s filings; 50/50 of Farquhar’s) — so the selling is programmatic/diversification, not necessarily a discretionary bearish signal, but the complete absence of any offsetting discretionary buy at a stock down ~60% from its high is notable. (FACT.)
• Magnitude (founders dominate): Co-founders Michael Cannon-Brookes and Scott Farquhar account for essentially all of the dollar volume. Parsed (conservative floor, the parser undercounts multi-line tables): Cannon-Brookes ≥896,835 shares (~$145M+) and Farquhar ≥386,246 shares (~$75M+) sold over the window; true totals are higher. Even at a 60%-off-high price, the founders kept selling on schedule and bought nothing. (FACT on direction; magnitude is a parsed floor — OPEN QUESTION on exact totals.)
• Note on Farquhar: Scott Farquhar stepped back from the co-CEO role (Cannon-Brookes is now sole CEO); his continued large 10b5-1 selling is consistent with a departing-founder diversification pattern rather than a fresh signal — but it is one-directional. (INTERPRETATION.)

Dual-class / voting reconciliation (DEF 14A, Oct-15-2025; record date Oct-8-2025)

• Class A = 1 vote; Class B = 10 votes (super-voting). (FACT.)
• Michael Cannon-Brookes: 48,024,933 Class B shares = 50.0% of Class B, 42.59% of total voting power.
• Scott Farquhar: 48,024,933 Class B shares = 50.0% of Class B, 42.59% of total voting power.
• Together: 96,049,866 Class B = 100% of Class B and 85.21% of total voting power (officers/directors as a group).
• Economic vs. control mismatch: total economic shares ≈ 262M (FY25 diluted 261.8M; public market data shows only ~160M Class A “shares outstanding” because it excludes the ~96M Class B). The two founders own ~96M of ~262M economic shares (~37% of the economics) yet control ~85% of the votes. Outside Class A holders have negligible governance power. (FACT.)

Insider verdict: Maximally unsupportive of a conviction-long read from the tape. Zero open-market buying at a deeply de-rated price; uninterrupted, programmatic founder selling of $200M+; and an entrenched dual-class structure (85% founder voting control) that insulates every capital-allocation and compensation decision from outside-shareholder discipline. The 10b5-1 framing softens the selling signal (it is scheduled diversification, not a panic), but the absence of any buy and the governance entrenchment are genuine negatives for the variant-perception and capital-allocation sections.

8. Changes and Headwinds — Last Two Years

The two years to mid-2026 reshaped Atlassian on five fronts: the deliberate end-of-life of its on-prem franchise, a string of strategically-mixed acquisitions, a governance simplification, a re-platforming of go-to-market around enterprise sales, and a full pivot to AI — all against a ~60% share de-rate from the ~$222 high.

1. Data Center end-of-life (Sept-2025) — the single biggest change. Atlassian announced EOL of its Data Center on-prem product (“Ascend”; final EOL March-2029), forcing its remaining large on-prem estate toward Cloud. Strategically coherent (Cloud is where AI, the Teamwork Graph, premium/enterprise tiers and ~83% gross margins live) but it injects multi-year revenue-recognition distortion: term-license revenue front-loads into FY26 (ASC 606), drawing down RPO/CRPO, and produces negative DC revenue and a total-growth trough in FY27 before FY28 reacceleration. Management is so concerned about investor confusion that it held a dedicated revenue-rec teach-in at Team '26. INTERPRETATION: correct long-term decision, but it converts FY26-27 reported financials into a puzzle and asks investors for patience precisely when sentiment is fragile.

2. Acquisitions — coherent core, questionable edge:

• Loom (~$975M, 2023): async video into Confluence/Teamwork — on-strategy, integrated, defensible.
• DX (~$720M, Nov-2025): developer-experience/engineering-analytics — clean adjacency to Jira/dev-tools, plausibly accretive to the enterprise motion.
• The Browser Company / BCNY (Arc & Dia browsers, $488M cash, Oct-2025): a consumer web-browser business. This is the strategic-fit head-scratcher. The rationale (“the browser is where work happens; an agentic browser as an AI surface”) is articulable but unproven, and it puts ~$0.5B of shareholder cash into a category with no obvious tie to Jira/Confluence economics and a history of consumer-browser monetization failure. INTERPRETATION: a yellow flag on capital-allocation discipline — a founder-controlled company spending half a billion on an adjacency that the market cannot underwrite. Buybacks meanwhile only offset SBC (share count still rising), so cash is going to M&A and dilution-mopping, not net return.

3. Governance / leadership: Co-founder Scott Farquhar stepped back from co-CEO in 2024; Mike Cannon-Brookes is now sole CEO. New CFO James Chuong (~Apr-2026) and CRO Brian Duffy (~2024). The two founders still control 85.2% of votes on ~37% of economics via 10-vote Class B super-shares — entrenchment is undiminished by the management refresh. INTERPRETATION: leadership depth improved (a real enterprise CRO, a fresh CFO), but accountability to outside shareholders did not.

4. PLG → enterprise sales re-platforming: The defining operational change. Quota-carrying reps expanded from ~117 to ~400; ASPs +22%; deal sizes and commitment durations up; $3M+ ARR cohort +54% YoY. Management is explicit that PLG runs alongside (not replaced by) the enterprise motion. INTERPRETATION: this is the engine of the up-market thesis and is working — but it raises the cost structure and lengthens sales cycles, and it is the lever that must keep working to deliver FY28’s reacceleration.

5. AI / Rovo + Teamwork Graph: Full-platform AI launch — Rovo (search/chat/agents) powered by the Teamwork Graph, monetized chiefly through the Teamwork Collection bundle plus consumption meters. Early traction strong (credits +20% MoM, 75% of F500 on Rovo, Rovo customers +2x ARR growth). Cloud price increases continue to layer in. INTERPRETATION: Atlassian has turned the existential “AI kills seat-based software” threat into a narrative offense — but monetization is still small and the thesis is unproven in the P&L.

VERDICT — net thesis-strengthening on substance, thesis-clouding on optics and discipline. The DC EOL, enterprise go-to-market, and AI platform are coherent moves that should make the business bigger, stickier, and higher-margin (GAAP operating profit is even guided to begin in FY27). But the changes also (a) make FY26-27 financials genuinely hard to read, (b) reveal a capital-allocation wobble (the BCNY browser deal) inside an 85%-vote-controlled structure, and © raise the execution bar. On balance the operating substance strengthens the long-term thesis; the optics and the browser deal weaken the near-term confidence — which is precisely why the stock is down ~60% while the business compounds.

9. Risk Analysis (Risk Matrix)

Top of the stack: Risks #1 (AI seat-compression) and #2 (Microsoft bundling) are the only two that can break the long-term thesis; #6 (SBC) is the one that quietly erodes per-share value every year regardless of operations; #4 (FY27 trough) is the near-term sentiment landmine. No identified risk of catastrophic/total loss — net-cash balance sheet, 83% gross margins, >99% high-end retention.

10. Valuation — Embedded Expectations

The setup. At $88.52 (2026-06-12), down ~60% from the ~$222 high, Atlassian carries a market cap of ~$23.2B and EV of ~$23.0B. On headline metrics it screens cheap — and on its own history, exceptionally so.

Multiples (FACT):

The contrarian hook is unmistakable: P/S at the 2.4th percentile and composite at the 15.6th mean the market has never valued a dollar of Atlassian revenue this cheaply as a public company. A 20%-grower (durable Cloud at ~29%) at ~13.6x forward non-GAAP earnings and ~4.3x EV/S is, on the tape, a value SaaS stock.

The tension — headline FCF vs. owner-FCF. This is the entire valuation argument:

• On headline FCF (~$1.4B, 27% margin), EV/FCF ≈ 16x — cheap for a 20%-grower.
• But SBC = $1,362M (26% of revenue), so owner-FCF (FCF − SBC) ≈ breakeven. Buybacks only mop up dilution; share count still rises. On an owner-economics basis, the company throws off ~zero distributable cash today and EV/owner-FCF is effectively not meaningful / very expensive. GAAP operating result is still a loss (FY25 −$130M). Rule of 40 = 52 headline / ~26 owner-adjusted. The cheap-looking stock is cheap only if you accept that SBC will scale down and headline FCF will convert to real owner cash. That conversion — guided to begin with GAAP operating profit in FY27 and an expectation that R&D need not grow with revenue — is the crux.

Embedded-expectations / reverse-DCF (ASSUMPTION-driven). To justify ~$23B EV, the market must be underwriting roughly:

• A revenue path through the FY27 trough into FY28 reacceleration — call it a ~17–19% revenue CAGR over the next 5 years (Cloud high-20s decelerating, DC negative then gone, AI layering in).
• A terminal owner-FCF margin (post-SBC) climbing from ~0% today toward the low-20s% as SBC/revenue normalizes from 26% toward a more typical mature-SaaS ~10–12% and enterprise scale delivers GAAP profitability.
• ~8–9% discount rate, ~3% terminal growth.

At today’s EV, the market is not pricing aggressive growth — ~4.3x EV/S on a 20% grower is undemanding. What it is pricing is a demand that SBC normalizes and owner-FCF inflects. In other words, the embedded expectation is not “growth re-accelerates forever” — it is “the business converts its high headline margins into real per-share cash as it matures.” That is a more achievable bar than the 2021 multiple implied, which is why the de-rate is rational rather than terminal.

Scenarios (5-yr, illustrative — ASSUMPTION):

• Bear: AI seat-compression bites + Microsoft bundling pressures net adds; revenue CAGR slips to ~10–12%, SBC stays elevated (~22–24% of revenue), owner-FCF stays near breakeven. The “cheap” P/S re-rates down further because there is no owner cash to value. Multiple compression on a falling-knife narrative.
• Base: FY27 troughs as guided, FY28 reaccelerates to high-teens/low-20s on Cloud + migration tail + AI; SBC grinds toward ~18% of revenue; GAAP op profit turns positive FY27 and builds; owner-FCF margin reaches low-teens by FY30. The headline multiple re-rates modestly as owner-economics become real. The de-rate proves to be a re-pricing, not a permanent impairment.
• Bull: AI is additive (consumption monetization + non-developer expansion outruns any seat compression); revenue CAGR ~20%+ sustained; SBC normalizes to ~12%; owner-FCF margin → low-20s. On real owner-FCF the current EV looks deeply cheap and the 2.4th-percentile P/S is a generational entry. This requires AI to be a tailwind, not a threat — the single swing factor.

Peer cross-read (FACT, public market data, reconcile to filings): TEAM fwd P/E ~13.6x at +20% durable Cloud growth sits cheaper or in line with slower peers — WDAY ~10.3x fwd at ~13% growth, CRM ~10.7x at ~13%, NOW ~20x at ~22%, MNDY ~14.4x at ~24% (closest pure work-mgmt peer), GTLB ~27x at ~23% (dev-tools peer). On EV/S, TEAM ~4.3x vs MNDY ~3.1x, GTLB ~4.7x, NOW ~7.5x. INTERPRETATION: against the closest growth-comparable peers (MNDY, GTLB), TEAM is not an obvious bargain on EV/S — it is fairly-to-attractively priced for its growth, with the kicker being its own-history percentile and the optionality on SBC normalization. The “cheapest ever” claim is true against itself, not against the cohort.

VERDICT (discussion, no target). Atlassian is cheap on the metric that flatters it (headline FCF, P/S percentile) and fair-to-expensive on the metric that matters (owner-FCF net of 26% SBC). The valuation is not pricing heroic growth; it is pricing a demand that the company finally converts software-grade margins into per-share cash. The embedded expectation is modest and achievable in the base case — which is what makes the setup interesting — but the 26% SBC and the FY27 trough mean the “cheap” screen is conditional, not unconditional.

11. Variant Perception

Consensus view. The Street largely accepts that (a) the FY26 print is accounting-inflated and FY27 will trough, (b) Cloud at ~29% with NDR >120% is healthy, and © the stock is “cheap” after a ~60% de-rate. The split is on the terminal question: is Atlassian a durable enterprise-platform compounder temporarily mispriced, or a seat-based software franchise about to be structurally disrupted by AI and bundling? The ~$143 average analyst target implies the sell-side leans constructive — but a 12.5% short interest of float says a meaningful cohort is betting the other way.

Strongest bull case. Atlassian is a misunderstood, de-risking compounder. The melting DC line is a deliberate amputation that converts a low-margin on-prem base into high-margin Cloud (93% landing in premium/enterprise, ARR +1.5-2x post-migration). The enterprise motion is working (ASP +22%, $3M cohort +54%, $1M+ cohort +39%). NDR is rising, not compressing — directly refuting the seat-disruption fear. AI is being weaponized offensively: the Teamwork Graph is a genuine context moat (demonstrably halves agentic-task cost), Rovo credits compound 20% MoM, 75% of the F500 are on it. GAAP profit begins FY27; R&D need not scale with revenue, so SBC/owner-FCF inflects. At the 2.4th percentile of its own P/S history, you are buying a 20% durable grower with a real moat at a trough multiple before an FY28 reacceleration. Falsified if: Cloud growth decelerates below ~20%, NDR breaks below 115%, or SBC fails to fall as a % of revenue for 4+ quarters.

Strongest bear case (what the 12.5% short is betting). Three legs: (1) AI seat-compression — the per-seat model is structurally short the AI transition; as agents do the work, human-seat growth stalls and the entire pricing architecture has to be rebuilt at lower revenue. (2) Deceleration — strip the DC pull-forward and the real trend is 34%→20% and falling; FY27’s negative DC line could trough below guidance and FY28 reacceleration may disappoint. (3) SBC / quality-of-earnings — 26% SBC means owner-FCF is ~breakeven; the “cheap” headline FCF is an illusion, share count keeps rising, and a founder-controlled board (85% votes) just spent $488M on a web browser. The stock isn’t cheap; it’s a value trap where the cash never reaches the owner. Falsified if: owner-FCF margin turns clearly positive and rising, NDR stays >120% through the AI transition, and SBC/revenue falls toward the mid-teens.

The 3–5 assumptions that matter most:

1. Does AI compress or expand seats? (The whole thesis.) Evidence so far — NDR rising, Jira seat expansion driving the reaccel, 2/3 non-developer users — leans expand. Unproven through a full cycle.
2. Does SBC normalize toward mid-teens % of revenue? Without it, owner-FCF never inflects and the “cheap” multiple is a mirage. Management’s “R&D need not grow with revenue” comment is the tell to track.
3. Is the FY27 trough as-guided, and does FY28 actually reaccelerate? The Cloud + migration-tail mechanics support it; execution and macro could break it.
4. Microsoft bundling pressure on net adds — the slow structural tail risk; watch high-end retention (>99% so far holds).
5. Capital-allocation discipline under 85% founder control — does the BCNY browser bet prove visionary or value-destructive?

The crux. This is a classic contrarian-value setup riding on one binary: is generative AI a tailwind or a wrecking ball for seat-based work software? Every other debate (FY27 trough, SBC, bundling) is second-order. The bull owns a 20% durable grower with a real context moat at a once-a-decade-cheap own-history multiple; the bear owns a decelerating seat franchise whose headline cash never reaches the owner, run by an unaccountable founder. The early evidence (rising NDR, seat-led Cloud reaccel, non-developer mix, Rovo cross-sell) tilts toward the bull — but the bear’s thesis is structural and slow, so the absence of disruption so far does not falsify it. The variant edge is that the market is pricing FY27’s optical trough and AI fear as if they were permanent, when the durable Cloud engine and migration tail argue they are temporary — provided SBC converts to owner cash.

12. Fact vs. Interpretation Table

13. Open Questions

1. Does AI compress or expand seats over a full cycle? The single most important unknown. Evidence so far (rising NDR, Jira-seat-led Cloud re-acceleration, ~2/3 non-developer users) leans toward expansion, but the bear thesis is structural and slow — absence of disruption to date does not falsify it. Watch net retention and large-account seat counts.
2. Does SBC normalize toward the mid-teens % of revenue? Without it, owner-FCF never inflects and the “cheap” multiple is a mirage. Management’s “R&D need not grow with revenue” comment is the tell; track SBC/revenue and the diluted share count quarter by quarter.
3. Is the FY2027 trough as-guided, and does FY2028 actually re-accelerate? The Cloud + Data-Center-migration mechanics support it, but a deeper/longer trough or a macro/tech-hiring downturn could break the FY28 reacceleration on which the thesis rests.
4. Does the BCNY/browser bet prove visionary or value-destructive? A $488M adjacency outside holders could not veto — the live test of capital-allocation discipline under 85% founder control.
5. Precise FY2025 product-line split (Cloud/Data Center/Marketplace %) and geographic mix — directionally Cloud ~63% / DC ~one-third / Marketplace ~4–5%, US ~45%; confirm exact figures against the FY2025 10-K MD&A.
6. Scott Farquhar’s exact current role post-co-CEO-step-back (board/special-advisor capacity) and the continuity plan around Cannon-Brookes as sole CEO with concentrated voting control.
7. A single consolidated net-revenue-retention figure — management has shifted toward ARR/cohort disclosure; the move away from a clean headline NRR is itself a disclosure-quality flag.

14. What Must Be True (Bull and Bear, with Falsification Tests)

For the BULL case to be right, the following must hold:

1. AI is a net tailwind, not a wrecking ball. Atlassian’s seat base keeps expanding (ex-migration) and consumption monetization (Rovo credits, agent runs) outgrows any per-seat erosion. Falsified if: net revenue retention breaks below ~115%, or large-account seat counts visibly decline, for two-plus consecutive quarters.
2. SBC converts to owner cash. SBC/revenue falls from 26% toward the mid-teens, GAAP operating profit emerges in FY2027 as guided, and the diluted share count actually stops rising and starts shrinking. Falsified if: SBC stays at 25%+ and the share count keeps climbing through FY2027 despite the larger buyback.
3. The FY2027 trough is temporary and FY2028 re-accelerates. The Data-Center migration tail and Cloud at high-20s carry total growth back to high-teens/low-20s in FY2028. Falsified if: FY2027 troughs materially below guidance or FY2028 fails to re-accelerate.

For the BEAR case to be right, the following must hold:

1. Seat-based work software is structurally short the AI transition. As agents do the work, human-seat growth stalls and the pricing architecture must be rebuilt at lower revenue. Falsified if: owner-FCF margin turns clearly positive and rising while NDR holds >120% through the AI transition.
2. The “cheap” multiple is a value trap. 26% SBC means owner-FCF stays near breakeven, the share count keeps rising, and founder-controlled capital allocation (browser deals, no buys) means the cash never reaches outside owners. Falsified if: SBC/revenue falls toward the mid-teens and the company begins returning net cash (shrinking share count) rather than recycling it to employees and adjacencies.
3. Microsoft bundling and modern-team defection slowly starve the franchise. Falsified if: high-end retention stays >99% and the enterprise $1M+/$3M+ cohorts keep compounding 35%+ through a Microsoft-Copilot/GitHub push.

The bull and bear share a single fulcrum: whether 26%-of-revenue stock compensation ever becomes per-share cash, and whether AI expands or compresses the seat. Both are observable within 3–6 quarters in the SBC/revenue ratio, the diluted share count, and net retention.

15. Source Appendix

The full source appendix is provided as a separate deliverable and stitched as Appendix B in the combined report.

APPENDIX A — Standard Diligence Questionnaire

Atlassian Corporation (NASDAQ: TEAM). Report date 2026-06-13. Supplemental to the memo. Fact/Interpretation/Assumption labels applied where it matters.

General

What thoughtful questions have other investors asked about this company? The dominant investor debates: (1) Is generative AI a tailwind or a wrecking ball for seat-based work software? — the binary the ~12.5%-of-float short interest is betting on. (2) Will 26%-of-revenue stock comp ever convert into per-share cash for outside owners, or is the headline FCF a mirage that accrues to employees? (3) Is the FY2027 revenue-growth trough (a self-inflicted product of the Data-Center end-of-life) temporary or the start of a deceleration? (4) Why is a founder-controlled company spending $488M on a consumer web browser at the same time the stock is down 60% and insiders are buying nothing? (5) Is Jira losing the future as modern teams adopt Linear/Notion, even as enterprise retention stays >99%?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? GAAP earnings are negative (operating loss every year since FY2023) and depressed structurally by SBC, not cyclically. Cash earnings (FCF ~$1.4B) are near a high on a headline basis but near breakeven on an owner basis. Reported revenue growth is at a deliberate, self-inflicted peak in FY2026 (Data-Center license pull-forward) heading into a guided FY2027 trough. (Interpretation.)

Driven by the external environment or internal actions? Predominantly internal: the Data-Center EOL, the PLG→enterprise GTM rebuild, the cloud-migration program, and the SBC budget are all management choices. External factors (tech-sector hiring, IT budgets, AI disruption) matter at the margin for seat growth. (Interpretation.)

How stable are revenues? Very stable and recurring — Cloud subscriptions plus Data-Center term licenses, 120%+ net retention, >99% retention in the $1M+ cohort, RPO +37% YoY to ~$4.0B. The instability is in reported growth optics (ASC 606 timing), not in the underlying revenue base. (Fact.)

Outlook for products/services? Cloud (~29% growth) and the enterprise/ITSM motion are healthy and expanding; Data Center is being deliberately euthanized (negative in FY2027); AI/Rovo is an early, inflecting option. (Fact + Interpretation.)

How big is this market — growing, shrinking, domestic or international? Management cites a re-cut serviceable market of ~$140B (up from ~$67B), framed against ~1 billion global knowledge workers (~900M non-technical). The directional growth is real (expansion beyond developers); the absolute number is a narrative to discount. ~45% of revenue is US, the balance international and under-penetrated. (Fact on mix; Interpretation on TAM.)

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More, on two fronts: Microsoft bundling (GitHub, Azure DevOps, Teams, Loop, Planner, Copilot) and an AI-enabled wave of design-led challengers (Linear, Notion) re-fragmenting the supply side. The 2022–24 rate reset thinned the herd, but AI lowered the cost to build a competing tool. (Interpretation.)

How profitable is the business (ROIC, ROE)? GAAP-unprofitable; ROE/ROIC are mechanically negative/meaningless (thin equity $879M, GAAP losses). The right lenses: ~83% gross margin, ~27% headline FCF margin (~1% owner-FCF margin), Rule of 40 = 52 headline / 26 owner. (Fact + Interpretation.)

How profitable is the industry — how many competitors, what barriers to entry? Best-in-class gross margins (80%+) but chronic fragmentation and one overwhelming bundled competitor. Barriers to entry are low (anyone can build a task tracker); barriers to displacement are high (embedded workflows, data, ecosystem) — the asymmetry that defines the moat. (Interpretation.)

Can the business be easily understood? Yes at the product level; the reported financials are temporarily hard to read because of the Data-Center EOL revenue-recognition distortion. (Interpretation.)

Can it be undermined by foreign low-cost labor? Not directly — it is IP/network-effect software, not labor-arbitrage-exposed. The relevant disruption vector is AI, not offshoring. (Interpretation.)

Do brands matter? Yes — and Atlassian’s developer brand is bifurcating: an asset in the enterprise installed base, a liability among modern/startup teams who view Jira as bloated and prefer Linear/Notion. (Interpretation.)

What is the nature of competition? Land-and-expand value selling, ecosystem lock-in, and increasingly AI capability — versus Microsoft’s bundle-and-undercut and the challengers’ UX. (Interpretation.)

Customers’ switching costs? Among the highest in software: years of embedded workflows, data, automations, permissions, integrations and third-party Marketplace apps. The Canal+ CTO’s “we’d be back to Excel files” captures it. This is the primary moat. (Fact/Interpretation; evidenced by >99% high-end retention.)

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The franchise itself — brand, 350k-customer install base, Marketplace ecosystem ($4B+ cumulative app sales), and the Teamwork-Graph data asset — is largely off-balance-sheet (internally generated). (Interpretation.)

Off-balance-sheet liabilities? Standard operating leases; no unusual structures identified. The largest economic liability not in operating income is the ongoing SBC dilution. (Interpretation.)

How conservative is the accounting? Revenue recognition is conservative (ratable Cloud; ASC 606 for term licenses). The aggressiveness is in how investors are steered to non-GAAP — the GAAP-to-non-GAAP bridge is almost entirely the SBC add-back, and management is shifting away from a single headline NRR toward selective cohort metrics (a disclosure-quality flag). (Interpretation.)

How CapEx-hungry is the business? Minimally — FY2025 capex $45M (~0.9% of revenue). Asset-light software. (Fact.)

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? ~$1.4B headline FCF (~breakeven owner-FCF). Uses: R&D-first (~45–50% of revenue), bolt-on M&A (Loom $975M, BCNY $488M, DX $720M), and buybacks that offset SBC. No dividend. (Fact.)

Significant acquisitions recently? Yes — BCNY/The Browser Company (Arc/Dia browsers, $488M, Oct-2025) and DX (developer-experience analytics, $720M, Nov-2025), funded partly by liquidating the entire $424M securities book; goodwill +~$1.0B in nine months. BCNY is an adjacency-stretch flag. (Fact + Interpretation.)

Buying back shares? Yes ($779M FY25; ~$1.5B in 9mo-FY26 at ~$100.51 avg) — but the diluted share count still rises ~1–1.5%/yr; the buyback is a dilution-mop, not net capital return. (Fact.)

Issuing large amounts of new shares to insiders? Yes — SBC $1.36B/yr (26% of revenue), the highest in the cohort; this is the core dilution issue. (Fact.)

Compensation policy of directors/management? Founder-centric and equity-heavy; the dual-class structure (Class B = 10 votes) gives the two co-founders 85.2% of votes on ~37% of economics. Outside holders cannot discipline pay or capital allocation. (Fact.)

Motivations of management? Long-term, product-and-mission-driven founders (Cannon-Brookes now sole CEO) — a genuine strength for vision and R&D, a genuine weakness for outside-shareholder accountability. The insider tape is 100% selling ($200M+ via 10b5-1), zero open-market buying into a 60% drawdown. (Fact + Interpretation.)

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No — a US-domestic C-corp (reincorporated Delaware 2022; founded Australia). Common stock, dual-class. (Fact.)

Dividend policy? None; no dividend has ever been declared. (Fact.)

How profitable is the business? See above — cash-profitable on a headline basis, near-breakeven on an owner basis, GAAP-unprofitable. (Fact/Interpretation.)

Is net income diverging from cash from operations? Yes, dramatically and by design: GAAP net loss (−$257M FY25) vs. OCF +$1.46B — the ~$1.7B gap is overwhelmingly the non-cash SBC add-back plus D&A and working-capital timing. The divergence is expected for the model but is precisely why owner-FCF (which charges SBC back) matters more than either GAAP NI or headline OCF. (Fact + Interpretation.)

Risks & Downside

What factors would cause the stock to decline? A net-retention crack below ~115% (AI seat-compression showing up); a deeper/longer FY2027 trough or failed FY2028 reacceleration; SBC staying elevated while the share count keeps rising (owner-FCF never inflects); a Microsoft-bundling step-change; further founder capital-allocation missteps. (Interpretation.)

Risk of a catastrophic loss? Low. ~83% gross margins, ~$1.1B cash vs. $1.0B modest fixed-rate debt, >99% high-end retention, and ~350k customers make a sudden impairment unlikely. The realistic downside is multiple compression on a value-trap narrative, not insolvency. (Interpretation.)

Chance of a total loss? Negligible on a multi-year horizon given the franchise, retention and balance sheet. (Interpretation.)

Recent News & Events

Has the business environment changed recently? Yes — the AI/agentic-software narrative re-rated the entire seat-based-SaaS cohort (TEAM −60% from its high), and Atlassian’s own Data-Center EOL (Sept-2025) reset its near-term revenue optics. (Fact.)

Significant acquisitions? BCNY (Oct-2025) and DX (Nov-2025) — see above. (Fact.)

Change in accounting policies? No policy change; the reporting noise is the ASC 606 effect of the Data-Center wind-down, not a policy shift. (Fact.)

Recent changes — new markets, facilities, management? Co-founder Scott Farquhar stepped back from co-CEO (2024); Cannon-Brookes is sole CEO. New CFO James Chuong (~Apr-2026), CRO Brian Duffy (~2024). Enterprise sales force expanded (quota carriers 117→~400); first GSI partnerships (Accenture, Deloitte, PwC). AI platform (Rovo / Teamwork Graph) launched and monetizing early. (Fact.)

APPENDIX B — Source Appendix

Atlassian Corporation (NASDAQ: TEAM). Report date 2026-06-13. Primary sources first. Third-party aggregator and public market data flagged and reconciled to filings where material.

Primary — SEC filings (EDGAR, CIK 0001650372)

1. Atlassian FY2025 Form 10-K (fiscal year ended June 30, 2025) — revenue, gross profit, operating loss, SBC, cash flow, segment/geographic disaggregation, risk factors. EDGAR.
2. Form 10-Q, quarter ended December 31, 2025 (filed 2026-02-06) — BCNY/DX acquisition accounting, marketable-securities liquidation, deferred-revenue swing.
3. Form 10-Q, quarter ended March 31, 2026 (filed 2026-05-01, accn 0001650372-26-000027) — Q3-FY26 revenue $1.787B, 9mo $4.806B, net loss, goodwill $2,303M, buyback (~$1.0B Q3 at $85.04 avg), RPO, debt.
4. DEF 14A proxy (filed 2026-10-15; record date 2025-10-08) — dual-class voting (Class B = 10 votes), founder ownership: Cannon-Brookes & Farquhar 96,049,866 Class B = 85.21% of total voting power; compensation structure.
5. Form 4 corpus (~165 filings, ~May-2025 → Jun-2026) — insider transactions; zero code-P open-market purchases; founder 10b5-1 selling.
6. 8-K filings — earnings releases (Q1/Q2/Q3 FY26); acquisition announcements (BCNY closed Oct-20-2025; DX closed Nov-10-2025); May-2024 senior-notes issuance ($500M @ 5.25% due 2029; $500M @ 5.50% due 2034).
7. EDGAR XBRL (RevenueFromContractWithCustomerExcludingAssessedTax, OperatingIncomeLoss, NetIncomeLoss, GrossProfit, ShareBasedCompensation, NetCashProvidedByUsedInOperatingActivities, PaymentsToAcquirePropertyPlantAndEquipment, PaymentsForRepurchaseOfCommonStock, WeightedAverageNumberOfDilutedSharesOutstanding, CashAndCashEquivalentsAtCarryingValue, StockholdersEquity, LongTermDebtNoncurrent) — five-year financial series, reconciled.

Primary — Management commentary (transcripts, third-party feed; treated as hypothesis, validated against filings)

8. Analyst/Investor Day, 2026-05-06 (~134K chars) — FY27 trough / FY28 reacceleration framing; SAM re-cut to ~$140B; Data Center “Ascend” EOL (March 2029); enterprise GTM (quota carriers 117→~400; GSI partnerships); cohort metrics ($1M+ +39%, $3M+ +54%); Rovo/Teamwork Graph; non-developer user mix.
9. Q3 FY2026 Earnings Call, 2026-04-30 — Cloud >$1.1B/+29%, RPO +37%, NDR >120% and rising, DC pull-forward (~$50M), ITSM competitive displacements.
10. Q2 FY2026 Earnings Call, 2026-02-05 — FY27 “25%+ non-GAAP operating margin” commitment; DC EOL dynamics.
11. Q1 FY2026 Earnings Call, 2025-10-30.
12. Selected conference presentations (BofA 2026-06-02; Jefferies 2026-05-27; Morgan Stanley 2026-03-04) — strategic framing.

Secondary — Industry, competitive and market data

13. Atlassian Developer Blog / Atlassian Marketplace (2025–26) — cumulative app sales ($4B+), app/partner counts, Forge platform terms.
14. Third-party competitive comparisons (2025–26): Linear vs. Jira (“How Linear became the new Jira”), Notion vs. Confluence, Monday/Asana/ClickUp — modern-team mindshare commentary.
15. Peer valuation comps (public market data, reconcile to filings): WDAY, CRM, NOW, MNDY (Monday.com), GTLB (GitLab) — forward P/E, EV/S, growth.

Aggregator / quantitative helpers (flagged, reconciled to filings)

16. third-party fundamentals + valuation_index (2026-06-12) — own-history valuation percentiles (composite 15.6th, P/S 2.4th, P/B 28.8th), snapshot, short interest (12.5% of float), description. News feed empty (count:0 — routine for clean US filer). Third-party signal, not evidence.
17. public market data (2026-06-12) — price $88.52, 52-week range $56.01–$222.59, EV/debt/cash. NOTE: sharesOutstanding (159.6M) excludes ~96M Class B super-voting shares — do not use for market cap; reconciled to ~262M economic shares via the 10-Q/XBRL diluted count.

Notes on sources

All figures are drawn from Atlassian’s SEC filings (10-K, 10-Q, DEF 14A, Form 4, 8-K), public earnings-call and investor-day transcripts, the Atlassian Marketplace/developer disclosures, and public market data. Third-party valuation-percentile and aggregator data are flagged where used and reconciled to filings; peer multiples are from public market data and should be reconciled to each peer’s filings.