TEAM · Atlassian Corporation · Stock research

Current Thesis

Atlassian spent the past year priced as a casualty of agentic coding: if agents write the tickets and the code, seat-based collaboration software loses its billing unit. The 2026-08-06 Q4 FY26 release attacked that premise with numbers — revenue $1,766M (+28% YoY), cloud revenue $1,213M (+31%), subscription ARR $6,606M at 2026-06-30 (+23% YoY) — and the market repriced it in one session, gapping the stock more than 30% on 2026-08-07 to a $149.07 close. The narrative leg on offer is monetized agents: Rovo, the MCP server and a free Jira Coding Agent turning Atlassian's project graph into the control plane that other coding agents call. The same release carries the counterweight in plain sight: FY27 total revenue growth targeted at roughly 13% against FY26's 26%, with Data Center revenue guided to fall about 17%.

Bullish and bearish views on Atlassian Corporation

The model's bull view on Atlassian Corporation (TEAM), in brief: Q4 FY26 beat on both lines (2026-08-06): revenue $1,766M, +28% YoY, against roughly $1.66B consensus per same-day press coverage; non-GAAP EPS $1.87 versus about $1.50 expected. The bear view: The FY27 guide halves the growth rate: approximately 13% total revenue growth versus 26% in FY26, with subscription ARR growth stepping down to about 18% from 23%. Both cases follow in full.

Bull Case

• Q4 FY26 beat on both lines (2026-08-06): revenue $1,766M, +28% YoY, against roughly $1.66B consensus per same-day press coverage; non-GAAP EPS $1.87 versus about $1.50 expected.
• Cloud is accelerating, not just holding: Q4 cloud revenue $1,213M, +31% YoY. That segment carries the FY27 target (~25.5% growth) and the Q1 FY27 target (~28.5%).
• GAAP economics inflected: Q4 FY26 GAAP operating margin 12% and GAAP diluted EPS $0.55, against a $(0.09) loss in the year-ago quarter.
• Cash conversion: Q4 FY26 free cash flow $475M (27% margin); FY26 free cash flow $1,319M (20% margin) on FY26 revenue of $6,572M.
• Growth is concentrated in large accounts: as of 2026-06-30, customers above $3M ARR grew over 50% YoY and customers above $5M ARR over 70% YoY, with management flagging an all-time record quarter for $1M+, $3M+ and $5M+ ACV deals.
• Agent adoption has usage metrics attached, not just slideware: the MCP server and Teamwork Graph CLI passed one million monthly active users, Rovo assisted actions rose over 50% QoQ, and more than 80% of the Fortune 500 use Rovo (all disclosed with the Q4 FY26 release).
• Distribution ahead of price: the Jira Coding Agent and integrations with Claude Code, Cursor and GitHub Copilot ship at no extra cost to paid Jira Cloud customers — a share-of-workflow play whose payoff appears in seats and ARR rather than in a new revenue line.
• Capital return and insider intent align: FY26 repurchases totalled $1.8B (19.1 million shares) with roughly $1.9B of authorization remaining, and the 2026-08-06 8-K disclosed CEO Mike Cannon-Brookes' intention to enter a Rule 10b5-1(c) trading plan for up to $250M of open-market Class A purchases, effective no earlier than 2026-08-07.

Bear Case

• The FY27 guide halves the growth rate: approximately 13% total revenue growth versus 26% in FY26, with subscription ARR growth stepping down to about 18% from 23%.
• Data Center is the mechanism. FY26 Data Center revenue was $1,830.9M of $6,572M total; FY27 guidance calls for that line to decline roughly 17%. A double-digit decline on that share of the base absorbs a meaningful part of cloud's contribution, and the prior migration pull-forward is the stated cause.
• Margins move the wrong way: FY27 non-GAAP operating margin targeted near 25% against 30% in FY26, while Q1 FY27 is guided near 28.5%. The step-down is back-half weighted, which pushes the burden of proof onto prints that do not land until 2027.
• GAAP remains thin: FY26 GAAP operating margin was 0.2% against 30% non-GAAP. Stock compensation does the work between those two numbers.
• Not every desk followed: TD Cowen maintained Hold on 2026-08-07 with a $145 target — below the $149.07 close — and UBS stayed Neutral at $160 on 2026-08-08.
• Overhead supply is real: the 2026-08-07 close sits 18.3% under the $182.36 52-week high. The band above current price is where the stock traded on the way down, and there is no scheduled company event to force absorption of it until late October.

Setup & Price Structure

Measured, as of the 2026-08-07 close: last price $149.07; 52-week high $182.36 (-18.3% below); three-month return +62.7%; RSI(14) 75.4.

One session has traded since the gap. No post-gap base exists yet, and a single bar defines no range. Contemporaneous coverage placed the 2026-08-07 move between 30% and 36% intraday; against the $149.07 close that implies a pre-print reference somewhere in the $110–115 area (inferred from the reported percentage move; the exact prior close is not asserted here). On that inference the gap is roughly $35 wide, and the first structural question is whether price constructs a shelf in its upper half or drifts back into it.

Positioning observables, stated as observables: RSI(14) at 75.4 sits at the extended end of its range. Twelve-plus sell-side target changes landed inside 48 hours — Oppenheimer $200, KeyBanc $185, Morgan Stanley $180, Wells Fargo $180, BTIG $180, Macquarie $170, Guggenheim $165, Truist $160, UBS $160, TD Cowen $145 — a coverage cluster compressed into two dates. Retail-facing outlets ran the "rally more than 63%" framing on 2026-08-07. The disclosed insider flow is buying intent, not selling: the CEO's announced $250M purchase plan, subject to a Rule 10b5-1(c) cooling-off period that means no Form 4 confirms an actual purchase for months. No equity or convertible issuance accompanied the print.

Life-cycle: ACCELERATING, dated to 2026-08-06 and 2026-08-07 — the print, the 30%+ gap, the same-week target cluster and the CEO purchase disclosure are new attention arriving simultaneously on a name that had been trading as a de-rating story. The label is fragile in one specific way: acceleration into a 75 RSI with no company catalyst scheduled until late October is acceleration with no forcing event behind it, and the flow that produced it can thin without any news to mark the turn.

Catalyst Calendar (next 30 days)

• 2026-08-31 — FY26 Form 10-K, statutory outside date. As a large accelerated filer with a June 30 fiscal year end, the annual report is due within 60 days (2026-08-29 falls on a Saturday, rolling to the next business day). The filing date is not company-confirmed. Resolves: segment detail and risk-factor language behind the ~17% FY27 Data Center decline, remaining performance obligations, and the share count after $1.8B of FY26 repurchases.
• No scheduled earnings, guidance or product event falls inside the 30-day window. The next binary is roughly twelve weeks out, which is itself a structural fact about how this leg has to trade.
• ~2026-10-29 (est.) — Q1 FY27 print. Tests the $1,705–1,715M revenue target, the ~28.5% cloud growth target and the ~28.5% non-GAAP operating margin target, and delivers the first observed Data Center decline against the ~17% FY27 path.
• ~2026-11-05 (est.) — earliest plausible Form 4 under the CEO trading plan. Rule 10b5-1(c) requires a cooling-off period for officers before trades may begin; on a plan adopted around 2026-08-07 the first filed purchase would not appear before early November.

What Would Change Our Mind

The structural break comes first: there is nothing built above the gap to lose yet, so the thesis rests entirely on whether the 2026-08-07 range becomes a shelf or a wick. A weekly close below $130 would return more than half the earnings gap and put the pre-print zone back in play, which reads as the market rejecting the re-rating rather than digesting it.

Fundamentally, the guide is the test. Cloud growth printing under the ~28.5% Q1 FY27 target at the estimated 2026-10-29 report, or subscription ARR tracking below the ~18% FY27 path, removes the offset that justifies absorbing a Data Center line guided down 17%. A cut to the ~25% FY27 non-GAAP operating margin target would say the free bundling of the Jira Coding Agent is costing more than it lands.

Two softer conditions worth watching without treating either as decisive. First, the 10b5-1 plan producing no filed Form 4 purchases by year-end 2026 would leave the CEO signal as an announcement only. Second, the theme flipping to SATURATED — the target cluster exhausted, coverage moving on, price grinding in the $150s while the broader software complex makes new highs — would mean the marginal buyer has already transacted.

Correlation Notes

• Seat-based enterprise SaaS re-rating. The move is legible as a read-through for the whole "AI eats per-seat software" cohort — MNDY, ASAN, HUBS, NOW, CRM. If Atlassian's print is treated as sector evidence rather than a company event, TEAM trades with that basket and gives back the idiosyncratic premium when the basket fades.
• Agentic coding stack. Product dependency runs directly through Claude Code, Cursor and GitHub Copilot, all named as integrations in the 2026-08-06 launch. Atlassian sits above them as the workflow layer, so it correlates with the agentic-coding narrative while carrying a distinct risk: those partners can become substitutes for the layer.
• Rates and duration. The gap landed on 2026-08-07, the same session that a -23,000 July payrolls print gutted September hike odds and drove the S&P 500 to a record with the Nasdaq up about 1.2%. Part of the day's magnitude was long-duration software beta; a hot inflation or payrolls print that re-prices hikes would pressure the multiple independent of anything Atlassian reports.
• Dev-tools peers. GTLB and the broader DevOps complex share the same customer budget line, so competitive commentary from those names in the October reporting cycle carries information for Atlassian's cloud growth trajectory.