Data as of 2026-09-14. The FY27 Q1 report in late October is a future event (the company has not announced a specific date; the timing here is estimated from its historical Q1 disclosure window), and we will update the readings against the verification calendar at the end of this note when it lands.
After the close on 6 August, Atlassian (TEAM) reported FY26 Q4 results (fiscal quarter ended 2026-06-30): revenue of $1,766M, up 28% year on year; Cloud revenue of $1,213M, up 31%, with the growth rate continuing to accelerate; adjusted EPS of $1.87, about 25% above the $1.50 consensus; and a full-year GAAP operating margin of 0.2%, the first positive full year. The next day (7 August) the stock closed up 35.3% at $149.07 (it was up as much as 32% in after-hours trading; this note uses the closing basis throughout). Going into the print it was down 32.1% year to date (from the 2025 year-end close of $162.14 to the 6 August close of $110.17), had at one point fallen as far as $57.15, roughly 65% below where the year started, and short interest stood at roughly 10-11% of the float, so short covering amplified the rebound. A month on, the re-rating has not stopped: since the start of September the stock has touched a 52-week high of $198.60, closing at $179.70 on 11 September for a market capitalisation of $45.60B, up roughly 63% from the pre-print $110.17.
Our core judgement is this: the beat, the AI quantification and the profitability inflection all survive verification, and the +35.3% has fundamental support. But the further advance over the following month has taken the current price into the lower end of the mainstream data providers' consensus target range (MarketBeat mean $183.81, stockanalysis mean $196.23), while the company's own FY27 guidance describes a different tempo — total revenue +13.0%, Cloud decelerating to +25.5%, GAAP operating margin 4.5%. The market has paid for "AI monetisation," yet Rovo still has no standalone revenue line, and the FY26 FCF margin fell 700 basis points. That gap between narrative and disclosure is the spine of this note, and the adjudication window is the FY27 Q1 report in late October.
Unpacking the triple catalyst: every leg of it is true
Start by setting out the composition of the +35.3%. The first leg is the size of the beat: revenue of $1,766M against consensus of roughly $1.66B, about 6% ahead; adjusted EPS of $1.87 against $1.50, 25% ahead. The second is the mix: Cloud revenue of $1,213M, up 31% year on year, accelerating past the full-year +28% level, which management attributed on the call to cross-sell and seat expansion, with the Data Center-to-Cloud migration contributing only mid-to-high single-digit percentage points. The bulk of the growth came from customers choosing to buy more, which directly falsifies the short narrative that cloud migration has peaked. The third is the inflection: a GAAP operating margin of 12% in Q4 alone and 0.2% for the full year (operating income of $10.4M), both positive for the first time.
The leading indicators point the same way: subscription ARR of $6.606B, up 23% year on year (management flagged that roughly one percentage point of that came from consolidating the DX acquisition); RPO of $4.8B, up 44%, far faster than revenue growth, with contracted but unrecognised large deals stacking up at an accelerating pace. The customer mix says the same thing: 57,334 customers with annualised Cloud ARR above $10K (+10.3%), 164 customers with annualised ARR above $3M and 69 above $5M, up 50% and 70% year on year respectively (earnings call), and more than 350,000 customers in total (10-K basis). Customer-count growth is running far below revenue growth, which says the engine is expansion within existing accounts, with new logos a supporting act. Put those numbers together and the directional verification is complete: enterprise customers are actively increasing what they buy from this platform.
Rovo: the halo is real, the revenue is missing
The AI-layer disclosure has to be read closely for what it actually says. The shareholder letter's wording is that Rovo adopters' ARR is growing more than 2x faster than non-adopters' — that is not Rovo's own revenue; it says customers who use Rovo spend more quickly across the whole platform. The accompanying depth metrics: more than 80% of the Fortune 500 use Rovo; Rovo-assisted actions grew more than 50% quarter over quarter; adopters complete 20% more Jira work items and create or edit 25% more Confluence pages than non-adopters; monthly active users of the MCP server and the Teamwork Graph CLI exceed 1 million, doubling in a single quarter. Both the breadth and the depth of adoption are rising, and we regard that halo effect as real.
But the monetisation structure means the market cannot see a dollar figure yet. Rovo is embedded inside Cloud subscriptions on a credit basis, and management was explicit on the call: the current vehicle for AI monetisation is subscription upgrades to the Teamwork and Service Collections, while direct overage-based consumption billing and the platform-level Flex model are "both early." More notable is a near-verbatim line in the 10-K risk factors: the strategy of offering AI tools to most Cloud customers free or at low prices may raise hosting costs without corresponding revenue. Read those three together and the conclusion is that Rovo is currently an on-ramp that pulls platform consumption, not a standalone product line. The "AI revenue" line the market most wants is one management has actively chosen not to give. The source materials contain no standalone Rovo revenue figure, and we do not estimate one here.
In fairness, the choice has competitive logic. Microsoft and Google are both bundling AI assistants into suites enterprises already pay for, so pricing Rovo separately would mean posting a standalone price tag in a market where rivals throw theirs in with the order. Embedding it free or cheaply buys adoption and data accumulation, and defends the installed base of more than 300,000 enterprise customers. The strategy itself is understandable, and so is the cost: as long as there is no dollar figure, "AI monetisation" is supported externally only by indirect evidence, and the ceiling on the valuation narrative is set by the disclosure basis — a switch management holds in its own hands.
The cost-side price has already been written into the guidance: FY27 GAAP gross margin guidance of roughly 84.5%, stepping down from the 85.0% guided for Q1 FY27. The company states in its annual report that it expects gross margin to decline slightly and that it will keep increasing investment in cloud infrastructure and AI. The higher the AI share, the thinner the gross margin, and that direction and Rovo's free-deployment strategy are mutually reinforcing.
FY27 guidance: the deceleration is written down on paper
The guidance repays a slow read. FY27 total revenue growth of roughly 13.0% is half the FY26 rate of +26%, dragged mainly by the deliberate migration that takes Data Center revenue down roughly 17.0%. Cloud revenue growth of roughly 25.5% decelerates from FY26's +28% and Q4's +31%. Subscription ARR growth of roughly 18.0% slows from FY26's +23%, with management describing itself as having built in an extra measure of prudence for the second half. The near-term Q1 FY27 guidance is relatively full: total revenue of $1,705-1,715M and Cloud up roughly 28.5%. Margins also need to be read with seasonality and cadence in mind: GAAP operating margin was 12% in Q4 alone, while Q1 FY27 is guided to only roughly 6.5% and the full year to roughly 4.5%. The direction of the inflection is confirmed, but the slope is far from a straight line, and extrapolating a single quarter's peak across the year would overstate the profit trajectory.
Cash flow is the negative signal the rally has covered up, and the boundaries belong with it. FY26 FCF was $1,319.1M at a 20.1% margin, down 700 basis points year on year (FY25 was 27.1%). Unpacked: operating cash flow of $1,353.1M, down 7.3%; cash severance from two rounds of restructuring during the year of $203.9M; and interest income falling from $112.3M to $69.7M (-38%), with those two items together explaining most of the decline. Capital expenditure was only $34.1M, roughly 0.5% of revenue, so the asset-light structure has not changed. In other words, the deterioration at the operating level is far smaller than the headline 700 basis points, and that is the boundary on the bear argument. Running it the other way, the quality of earnings in this inflection year still depends on clearing the restructuring and on interest income stabilising, and management has given no FY27 FCF figure, which is the boundary on the bull argument.
Stock-based compensation and buybacks belong in the same entry. SBC of $1,605.1M is 24.4% of revenue, at the high end among software companies of similar size; the offset is $1.81B of buybacks (19.1M shares at an average of roughly $94, under 60% of the current price), diluted share count fell on a net basis in FY26, and the $2.5B authorisation added in October 2025 is roughly 5.5% of the current market capitalisation. The buyback is genuinely doing work, and it was executed at low prices. But what it consumes is precisely FCF, and as long as the SBC ratio does not come down, this hedging machine has to keep running.
Put the pricing back against those numbers: a market capitalisation of $45.60B on 11 September against FY26 revenue of $6,572.3M is roughly 6.9x sales (TTM); if FY27 revenue lands near $7.4B on the guided +13%, the forward multiple is roughly 6.1x. The multiple itself is not expensive among software peers. What is expensive is the position. Sell-side ratings stand at 25 Buy, 4 Hold and 1 Sell, mainstream data providers' consensus targets run between $183.81 (MarketBeat) and $196.23 (stockanalysis), and at $179.70 the current price has entered the lower end of that range with upside compressed to single-digit percentages. Even the most aggressive raises since late August (BofA $210, BTIG $230) sit only 10-30% above the current price. Sell-side optimism is largely written into the price, and the next leg up needs a new disclosure basis rather than a repetition of the old story.
The counter to the counter: the room left by RPO and prudent guidance
The two strongest objections to the "already priced" judgement are the following, set out as they stand. The first is RPO +44%: contracted value is growing 18 percentage points faster than revenue, and if those large deals convert normally over 12-18 months, the +13% in the FY27 guidance is most likely only a floor; management's self-described prudence on the ARR guidance also leaves room for a beat. The second is competitive position: the 10-K lists AI-native competitors as a distinct risk for the first time, and the boundary friction between tools like GitHub Copilot and the Jira ecosystem inside development workflows is a long-run variable. But the switching costs created by more than 300,000 enterprise customers and the accumulation of work data (Teamwork Graph) show no public evidence of loosening in the near term. Our judgement is that these two change the thickness of the downside protection without changing the main logic that a second ignition requires a dollar-denominated Rovo disclosure: RPO conversion supports delivery against the guidance, and it cannot light a new valuation narrative.
Two data gaps, self-reported. The company's annual report gives a continuous series only for the count of customers above the $10K tier in its key metrics; the $3M and $5M tiers cited above come from the earnings call and carry only year-on-year growth rates (+50%/+70%) with no continuous series of absolute figures, so changes in the large-customer mix can only be inferred obliquely from RPO and management's qualitative remarks. And the Rovo adopter-versus-non-adopter ARR comparison comes without a base or a grouping methodology, so selection bias — high-growth customers being the ones that adopt new features earliest anyway — cannot be ruled out with public data, which is why we record the 2x statement as halo evidence rather than causal evidence.
Verification calendar
| Timing | Event | How to read it |
|---|---|---|
| Late October (estimated from the company's historical disclosure window) | FY27 Q1 results | Total revenue against the $1,705-1,715M guidance; Cloud against the +28.5% line; whether Rovo gets a dollar-level disclosure basis for the first time (the confirming switch); whether overage billing and the Flex model move from "early" to a disclosable stage |
| Shareholder letter, same event | Rovo depth metrics | Whether quarter-over-quarter growth in assisted actions holds the +50% order of magnitude; whether the "adopter ARR 2x" statement continues — dropping the disclosure would itself be a negative signal |
| Each quarter of FY27 | Migration and conversion | The quarterly cadence within the full-year Data Center revenue guidance of -17.0%; the slope at which RPO +44% converts into revenue |
| Ongoing | AI-native competition | Expansion by GitHub Copilot and others into project management and knowledge collaboration features (via public channels such as GitHub Universe); a verifiable large-customer displacement case would escalate this to a first-order variable |
| Ongoing | Buyback execution | The degree to which the pace of execution against the $2.5B authorisation offsets the 24.4% stock-based compensation ratio (tracked with the quarterly cash flow statements) |
To draw it together: the Q4 triple catalyst (the beat, Cloud accelerating to +31%, GAAP turning positive for the first time) is real and has been fully rewarded by the market, and the directional verification is complete. But what the FY27 guidance describes is a decelerating structure of +13% total revenue and +25.5% Cloud growth; the 700 basis point drop in FCF margin and stock-based compensation at 24.4% of revenue are two real entries sitting underneath the rally; and the "AI monetisation" the market has paid for still has no dollar figure attached. With the price inside the lower end of the consensus target range and upside compressed to single-digit percentages, this stock has entered a phase of waiting for a disclosure basis.
The next observation point is the FY27 Q1 report in late October and the shareholder letter that accompanies it. Delivery against the 28.5% Cloud line, and the presence or absence of a Rovo disclosure basis: those two readings decide whether this re-rating continues or digests. If a formal product announcement on Rovo's billing model appears mid-quarter, re-read early; otherwise, let the guidance and the disclosures speak for themselves.
This note sets out Lunartulip Lab's research views, compiled and analysed from public information. It does not constitute investment advice. Markets carry risk; decisions must be made independently.