Atlassian TEAM

Reference table

1. What they actually sell — and the disclosure problem

Atlassian's segment reporting is, frankly, thin. The FY25 product split in this dataset shows just two lines:

That is the entire disclosure. FY24 additionally broke out a $177.2M "Maintenance" line and FY23 a $399.7M one — the legacy on-premise maintenance stream that has been running off as the company completed its cloud migration. By FY25 it is gone from the disclosure entirely.

The practical consequence: this file gives us no way to separate cloud subscription revenue from Data Center term licenses, no seat counts, no net revenue retention, and no product-level split between Jira, Confluence and the rest. That is a material gap and it is worth stating plainly, because it is exactly the data you would want in order to adjudicate the AI-disruption debate. We are underwriting a consolidated number.

Geography (FY25) is better disclosed:

Growth is broad-based across all three regions — a genuine positive, and evidence against the idea that the deceleration is a single-market problem. EMEA actually grew fastest.

2. The numbers that are not in dispute

Revenue (fiscal year ends 30 June):

FY26 is nearly complete: Q1 $1.433B, Q2 $1.586B, Q3 $1.787B (actuals) plus the $1.663B Q4 consensus = $6.469B, which matches the $6.470B consensus revenue estimate exactly. Note that FY26 growth accelerated versus FY25 — from +19.7% to +24.0%. Quarter-by-quarter year-over-year: Q1 +20.6%, Q2 +23.3%, Q3 +31.7%. The trend within the year is up, not down. That is the strongest single fact in the bull case and it deserves to be stated before the bear case gets its turn.

Profitability — the GAAP-versus-non-GAAP fork. This is where every valuation argument about Atlassian lives:

Atlassian has posted a GAAP net loss in every one of the six fiscal years in this dataset (FY20 through FY25), while generating positive and growing free cash flow throughout. The bridge is almost entirely stock compensation. Every forward P/E quoted for this company — including ours — is a non-GAAP P/E that excludes a $1.36B annual expense. We say so rather than hiding it, and we note the mitigant: the buyback has been large enough to keep the diluted share count essentially flat (249.7M FY21 → 253.3M FY22 → 256.3M FY23 → 259.1M FY24 → 261.8M FY25 → 261.0M in Q3 FY26). Over five years that is +4.5% cumulative dilution against 149% cumulative revenue growth. The stock compensation is a real economic cost; it is not, on this record, a dilution machine.

Balance sheet (FY25): cash $2.513B + short-term investments $0.424B = $2.937B against total debt of $1.239B → net cash of $1.698B (my computation). Note: the vendor's netDebt field reports -$1.274B because it nets only cash and excludes short-term investments. I use the fuller figure and label it.

One vendor field I reject: km_ttm reports enterpriseValueTTM of $27.349B against a market cap of $27.242B — i.e., it treats Atlassian as carrying $107M of net debt when the balance sheet shows $1.70B of net cash. That is the enterprise-value error pattern this house has seen repeatedly in this feed. My computed EV is ~$25.54B, and every EV multiple in this dive uses that number.

3. The three yellow flags in the financials

Honesty requires flagging these before the valuation section, because two of them cut against the bull case.

(a) TTM free cash flow appears to be going backwards. FY25 free cash flow was $1.416B. The vendor's TTM price-to-free-cash-flow ratio of 22.61x against a $27.242B market cap implies TTM free cash flow of ~$1.205B — roughly 15% below FY25, during a year in which revenue grew 24%. The file contains no quarterly cash-flow statements, so I cannot decompose this (timing of collections, tax payments, the FY25 comparison base). I flag it as unexplained and material. Growing revenue 24% while free cash flow declines 15% is the kind of divergence that either resolves benignly in one quarter or turns out to be the story.

(b) The Q4 consensus implies a sequential revenue decline. Q3 FY26 revenue was $1.787B. The Q4 consensus is $1.663B — a 7.0% sequential drop. Subscription businesses do not normally do that. The most likely benign explanation is that Q3 contained lumpy upfront term-license revenue (Atlassian's Data Center product recognises a portion of multi-year term licences at the point of sale, which makes quarters uneven); note that Q3's +31.7% year-over-year growth was itself well above the Q1/Q2 pace of +20.6%/+23.3%, which is consistent with that reading. But this is a hypothesis, not a fact in the file, and I label it as such. The alternative reading — that the Q4 estimate simply has not been refreshed — is equally available. Either way, do not treat Q3's 31.7% as the run-rate.

(c) One historical estimate row is internally inconsistent. The FY23 (2023-06-29) estimate row carries epsAvg of +$1.758 alongside netIncomeAvg of -$504.9M. Positive EPS with negative net income is impossible on a single basis; the row mixes GAAP and non-GAAP fields. It does not affect any forward number used here, but it is a reminder that the estimate series is not internally uniform.

What I did NOT find corrupted, having checked: the most recent quarter (inc_q[0], Q3 FY26) is clean — net income -$98.4M against 260.965M shares gives -$0.377, matching the reported -$0.38 EPS; interest expense and share counts do not duplicate the prior quarter; the earnings-calendar actuals ($1.04, $1.22, $1.75) reconcile to the FY26 consensus of $5.48 once the $1.49 Q4 estimate is added ($5.50). The GAAP-negative / non-GAAP-positive split here is a real accounting fact, not a data error.

4. The knowledge base — the panel is arguing AGAINST this name

This is the most important section in the dive, and it is unusual: of five tagged claims across five tracked sources, three are explicitly bearish and only one is substantively bullish.

The bear lane (three independent sources, six months, escalating):

• no_priors, 2026-02-19 (bearish, conviction 55): "Ticketing/project-management systems (Jira-type) are somewhat at risk as AI reshapes engineering workflows." — the mildest of the three, and it names Atlassian's flagship product category directly.
• a16z, 2026-03-20 (bearish, conviction 72, entities: monday.com, Atlassian): "'Beta' SaaS that makes customers similar to peers will struggle as AI/vibe-coding collapses switching costs; commodity software faces pressure." — this is the sharpest formulation of the structural case, and it names Atlassian explicitly.
• All-In, 2026-03-27 (bearish, conviction 65, 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." — note this one is descriptive rather than predictive; it observes the de-rating that has already happened.

The bull lane (one substantive claim):

• cognitive_revolution, 2025-11-20 (bullish, conviction 70, entity: Atlassian): "Atlassian's 20-year 'open by default' data and permissioned enterprise search are structural advantages for deploying useful enterprise AI agents."

One older, tangential claim:

• invest_like_the_best, 2022-07-26 (bullish, conviction 65, entities: Atlassian, ServiceNow, Salesforce, Workday, Zoom): "A new tier of large tech acquirers (Atlassian, ServiceNow, Salesforce, etc.) has emerged, widening M&A exit options for entrepreneurs." — this is a claim about the venture exit market, not about Atlassian as an investment. Counted, but carrying essentially no weight.

Read. The panel's net position on TEAM is negative, recent, and specific. Crucially, the bull claim and the bear claims are not really in contradiction — cognitive_revolution argues Atlassian has the data to build good AI agents; a16z and no_priors argue that the pricing model and switching costs around that data are what erode. Both can be true, and the combination is exactly a fairly-valued stock.

A note on search hygiene. A naive text search for this ticker returns fourteen hits, nine of which are false positives generated by the ordinary English word "confluence" — Ray Dalio on a "confluence of debt/money problems", Darius Dale on a "confluence" of macro factors, Money of Mine on a "confluence of positive factors" in precious metals. None of these concern Atlassian's Confluence product. Matching on product names rather than company names is how a KB count gets inflated from 5 to 14. Ours is 5.

5. Competitive position and moat

Atlassian's moat is the classic bottoms-up software moat: the product is adopted by individual engineering teams without a procurement process, spreads laterally through an organisation, and then becomes the system of record for how work is tracked. Once a company's entire project history, sprint structure, permissions model and integration surface live in Jira and Confluence, the cost of moving is organisational rather than financial.

The evidence in this file that the moat is still working: 83.9% TTM gross margin (up, not down — FY25 gross margin was 82.8%), a negative 22.4-day cash-conversion cycle (customers pay before Atlassian pays its own bills, the hallmark of pricing power in subscription software), $2.243B of deferred revenue at FY25 (+22.9% year over year, growing slightly faster than revenue), and broad-based geographic growth.

The evidence the moat is being tested: the bear lane above, and the plain fact that the market has removed 44% of the equity value.

Our assessment: the moat is real but narrowing at the edges. Switching costs built on "all our history is in here" are strong against a competing tool and weak against a change in how work itself is organised. The a16z formulation — that AI collapses switching costs specifically for software that makes customers similar to their peers — is the most intellectually serious version of the bear case, and nothing in this dataset refutes it.

6. Valuation — priced in or room?

At $103.71 (market cap $27.24B, net cash $1.70B, EV ~$25.54B):

Note the coverage cliff: FY26 and FY27 estimates rest on 16-17 analysts; FY28 on 7 and FY29 on 3, with an FY29 EPS range of -$0.10 to $13.49. Treat anything beyond FY28 as decoration.

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

At $103.71 — 17.1x FY27E non-GAAP EPS — the price embeds roughly the following falsifiable claims:

• Revenue compounds at ≥13% through FY27 and ≥15% through FY28 (consensus: +13.4% FY27, +15.9% FY28), i.e. the deceleration from 24% stops at 13% rather than continuing to single digits. Most fragile assumption in the price — see below.
• Non-GAAP EPS reaches ~$7.18 by FY28 (consensus, 7 analysts), requiring roughly 15% annual EPS growth from the $5.48 FY26 base, which in turn requires operating leverage on a cost base where R&D alone is 51.9% of revenue.
• Gross margin holds above 82% (FY25: 82.8%; TTM: 83.9%) — i.e. no pricing concession forced by AI-native competitors or by customers renegotiating seat counts downward.
• Free cash flow re-accelerates from the current ~$1.20B TTM back above the $1.42B FY25 level, since a 14.5x FY28 earnings multiple on a company that converts roughly a quarter of revenue to cash is only defensible if that conversion holds.
• The market keeps paying ~17x forward — i.e. no further de-rating, despite the fact that the same market has already cut the multiple by 44% in twelve months and three tracked sources argue the process is unfinished.

The single most fragile assumption: that 13.4% is the trough rather than a waypoint. Everything else in the list follows from it. If FY27 revenue growth guides to single digits on 2026-08-06, the 17x multiple is not a floor.

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

Expected return over the 6-24 month horizon ≈ EPS growth + multiple drift + shareholder yield.

• EPS growth: consensus +10.6% FY26→FY27 ($5.48 → $6.06), then +18.4% FY27→FY28. Call it ~14% annualised.
• Multiple drift: our base case assumes the market pays 18x forward at the horizon versus 17.1x today — a marginal expansion, and we want to be explicit that this is the weakest leg of the bridge. We are not forecasting a re-rating back toward the 29x the stock carried on FY25 earnings; we are assuming the de-rating stops. The mechanical rolldown from 18.9x FY26E to 17.1x FY27E at a constant price is not re-rating and we do not count it.
• Shareholder yield: no dividend. Buybacks ($779M in FY25, 2.9% of market cap) are consumed by offsetting the $1.36B stock-compensation charge, so the net shareholder yield is approximately zero — the buyback keeps the share count flat rather than shrinking it. This matters: it means essentially all of the expected return must come from earnings growth and multiple, with no cushion underneath.

Our base case is therefore almost entirely an earnings-growth story with a flat multiple. $109 = 18x FY27E EPS of $6.06. That is a +5.2% return from $103.71 over the horizon. A thesis that needs multiple expansion to work is a fragile thesis, and we are deliberately not building one.

Scenario bands (stated arithmetic, no DCF):

• Bear ~$67 — 11x FY27E EPS of $6.06. The de-rating continues because FY27 guidance comes in below 13%, the a16z thesis starts showing up in net revenue retention, and the market applies a mature-software multiple. -35% from spot.
• Base ~$109 — 18x FY27E EPS of $6.06. Growth stabilises around the consensus 13-16%, free cash flow recovers toward $1.4B, the multiple holds. +5% from spot.
• Bull ~$158 — 22x FY28E EPS of $7.18. The AI-agent thesis (cognitive_revolution's claim) converts into a visible new revenue line, growth re-accelerates above consensus, and the multiple partially recovers. +52% from spot.

Base sits ~5% above spot and 19% below the $134.89 street consensus. That gap is the dive's actual conclusion: we think the sell-side is paying for a re-rating we do not underwrite.

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

• We differ from the street, not from the panel. The street's $134.89 consensus (29 buy, 14 hold, zero sell) implies roughly 22x FY27E — a multiple expansion from today's 17.1x. We see no catalyst in this dataset that produces that expansion, and three tracked sources arguing for the opposite. Our variant perception is against the sell-side's implied re-rating, and it is the only edge we claim here.
• We do NOT have a variant perception on the fundamentals. Our revenue and EPS assumptions are consensus. Per house rule, a name where our view of the business ≈ consensus defaults toward Hold/Watch, and that is precisely where this lands. We are not going to manufacture conviction from a $5 gap to fair value.
• Positive surprise that would force a repricing: FY27 revenue guidance on 2026-08-06 materially above the 13.4% consensus, ideally accompanied by a separately disclosed AI/agent revenue line. That single event would falsify the a16z and no_priors claims in one move and justify the bull case. Watchable, dated, binary.
• Negative surprise that would force a repricing: FY27 guidance in single digits, or any disclosure that net revenue retention has fallen below ~110%, or a second consecutive quarter of free cash flow below the FY25 run-rate. The first two are the direct fingerprints of the switching-cost thesis; the third is the flag already raised in section 3(a).

7. Technicals — a base attempt, not a confirmed turn

• Price $103.71, -44.1% from the $185.69 52-week high, +81.5% off the $57.15 52-week low. Maximum drawdown from peak: -77.4%.
• Above a rising 50-DMA ($90.88) by 14.1% — near-term momentum is genuinely positive. Below a falling 200-DMA ($108.68) by 4.6% — the primary trend is still broken. The 200-DMA at ~$109 is the level, and it sits almost exactly at our base fair value. That coincidence is not meaningful in itself, but it does mean the technical and fundamental "prove it" lines are the same number.
• RSI 62.3 — neither oversold nor extended. MACD +3.60 — positive. 3-month return +16.7% against SPY's +5.1% and QQQ's +3.8%: the stock has been outperforming for a quarter.
• 12-month return -45.9% against SPY's +19.9% and QQQ's +23.9% — a 66-point relative deficit. This is a repair job in its early innings.
• Minor data inconsistency to note: the technical block reports a 52-week range of $57.15-$185.69 while the profile block reports $56.01-$189.69. The discrepancy is small and does not change any conclusion, but the two ranges in this file do not agree.

8. Insider activity

The available Form 4 record is thin and one-directional but small in size: Chief Revenue Officer Brian Duffy sold across four dates in June 2026 — 1,813 shares at $92.79 (06-10), 1,432 at $88.79 (06-11), 3,000 at $89.71 (06-12), plus an earlier 06-08 transaction — reducing his holding from 235,390 to 227,691 over the window. That is roughly 7,700 shares, under $700K, and about 3% of his position. No purchases appear in the record.

Read: negligible. This is routine, small-scale officer selling, well below the threshold at which insider activity carries information, and it occurred at prices ($88-93) below today's $103.71. It is neither a warning nor a comfort. We note it for completeness and weight it at approximately zero.

9. Verdict, kill-criteria and flip conditions

Watch. Atlassian is a high-quality, cash-generative, net-cash software franchise trading at a fair — not cheap — 17.1x FY27E non-GAAP earnings after a 44% de-rating. We decline to buy it today for three specific reasons, each of which is falsifiable:

1. Consensus itself models growth halving from +24.0% (FY26E) to +13.4% (FY27E). Our base fair value of $109 is only 5% above spot precisely because that deceleration is already the shared assumption.

2. The only tracked-expert lane on this name is net bearish, and the bear case (AI collapsing the switching costs of seat-priced workflow software) is structural, recent, argued independently by three sources, and unfalsified by anything in this dataset.

3. TTM free cash flow appears to be running ~15% below FY25 while revenue grew 24% — an unexplained divergence we cannot decompose with the data available, and one we are unwilling to underwrite blind.

Against that: earnings are in two days, and the FY27 guide is exactly the event that resolves point 1 and largely resolves point 2. Buying ahead of it would be paying for information we can have on Thursday for free.

Pre-registered ENTRY TRIGGERS (any one converts Watch → Stage-In, ≤1.5% starter):

• FY27 revenue guidance on 2026-08-06 above ~16% — materially better than the 13.4% consensus, which would falsify the deceleration assumption embedded in the price.
• A separately disclosed AI/agent revenue line of meaningful scale, converting the cognitive_revolution claim from thesis to line item.
• Price below ~$88 (roughly the 50-DMA and ~14.5x FY27E) without a fundamental deterioration — at that multiple the bear case is substantially pre-paid and the risk/reward inverts.
• Two consecutive quarters of free cash flow above the FY25 run-rate, resolving the section-3(a) flag.

Pre-registered AVOID criteria (any one converts Watch → Avoid):

• FY27 revenue guidance in single digits — the bear case moving from thesis to fact.
• Net revenue retention disclosed below ~110%, or any commentary indicating seat-count contraction at existing customers.
• Gross margin below 80% — evidence of pricing concession.
• The diluted share count resuming its climb (i.e. the buyback no longer absorbing the $1.36B stock-compensation charge), which would convert a manageable expense into genuine dilution.
• A third consecutive quarter of declining free cash flow.

Where TEAM fits in the Synthos Framework Portfolio. It does not, today. If an entry trigger fires, it belongs in the enterprise-software sleeve at ≤1.5% as a valuation-and-recovery position, explicitly sized small because the KB lane is net negative — we do not carry core weight in names our own tracked sources argue against. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $103.71.

Single biggest risk: that the 13.4% FY27 consensus growth rate is a waypoint rather than a trough. Every other number in this dive — the 17x multiple, the $109 base case, the free-cash-flow recovery — is downstream of it.

Most fragile assumption in the price: that the market keeps paying ~17x forward earnings for a business whose growth rate is expected to halve, in a category three tracked sources have independently identified as AI-exposed. The multiple, not the earnings, is the thing that breaks first.

Provenance & disclosures

• Traceability: 5 tagged KB claims across 5 tracked sources (breadth 5, net conviction negative-moderate). Bear lane: no_priors 2026-02-19 (conviction 55), a16z 2026-03-20 (72), All-In 2026-03-27 (65). Bull lane: cognitive_revolution 2025-11-20 (70). Tangential: invest_like_the_best 2022-07-26 (65, an M&A-market claim, near-zero weight). Nine additional text matches were excluded as false positives generated by the ordinary English word "confluence" rather than the Confluence product.
• Data as-of: fundamentals 2026-03-31 (FY26 Q3; fiscal year ends 30 June) · estimates 2026-08-04 · prices 2026-08-04 ($103.71, 50-DMA $90.88, 200-DMA $108.68, RSI 62.3) · KB claims 2026-08-04. Sole market-data source: the Synthos vendor file for this ticker. No figure in this dive is drawn from memory or from any other source.
• Data gaps explicitly noted: no product-level revenue split (94.5% of FY25 revenue sits in one "License and Service" line); no seat counts, net revenue retention, or cloud-versus-Data-Center split; no quarterly cash-flow statements (so the TTM free-cash-flow decline flagged in section 3(a) cannot be decomposed); FY28-FY29 estimates rest on 7 and 3 analysts respectively with an FY29 EPS range of -$0.10 to $13.49.
• Data rejected or corrected: (1) the vendor's enterpriseValueTTM of $27.349B implies net debt where the balance sheet shows $1.70B of net cash — all EV multiples here use my computed $25.54B; (2) the vendor's netDebt field of -$1.274B nets only cash and excludes $424M of short-term investments — I use $1.698B and label it as my computation; (3) the FY23 estimate row carries positive EPS alongside negative net income (a GAAP/non-GAAP mix) and is unused; (4) the profile and technical blocks report different 52-week ranges ($56.01-$189.69 versus $57.15-$185.69). The most recent quarter (inc_q[0]) was checked for the usual failure modes and is internally consistent.
• Non-GAAP caveat: every forward P/E in this dive uses consensus non-GAAP EPS, which excludes a stock-compensation charge running 26.1% of revenue ($1.362B in FY25). GAAP EPS was -$0.98 in FY25 and has been negative in all six fiscal years shown. This is disclosed rather than buried because it is the single largest difference between the headline valuation and the economic one.
• Fair-value caveat: the $67 / $109 / $158 anchors are scenario multiples applied to consensus non-GAAP EPS, not a discounted cash-flow model. The base case assumes the multiple holds at ~18x; it does not assume a re-rating.
• Timing caveat: this dive is written two days before the FY26 Q4 print (2026-08-06), deliberately, because the FY27 revenue guide is the event that resolves the central question. The verdict is Watch partly because that information is imminent.
• Not investment advice. Independent research, educational and informational only, never personalised.
• Version: 2026-08-04-full.