Platforms

Atlassian's index finds AI writes 70% of merged code, and the work left is what it sells

The first quarterly AI SDLC Index shows code generation getting dramatically cheaper while the scarce work moved to human review, planning and documentation, the products Atlassian sells. It lands as investors ask whether AI deletes the company's per-seat business, and it runs on data from DX, a business Atlassian bought.

Richard TangRichard Tangthree-minute read
Share
Mike Cannon-Brookes, co-founder and chief executive of Atlassian, photographed at the company's Team '26 event
Mike Cannon-Brookes, Atlassian's co-founder and chief executive, introduced the company's first AI SDLC Index personally this week.

Atlassian−2.63% — Atlassian, down 2.63 percent as of its last quote, the software group behind Jira and Confluence (ticker TEAM), published the first edition of its AI SDLC Index (SDLC: software development life cycle) this week and promised a quarterly update, giving the market a recurring read on what AI does to software work 12. Across more than 500 engineering organizations, AI-written code reached 70% of merged code in October, up from 49% in April 1. Generation is now the cheap part of software; the costly work is moving into review, planning and documentation 1.

Cheap code pushed the pressure into human review

AI spend per unit of output fell by more than a quarter in October 1. Survivorship, the share of AI-written code still present in later commits, recovered to 68% 1. Human review of pull requests, the checkpoints where engineers approve colleagues' code before it merges, grew at its fastest month-over-month rate to date 1. Engineers still spend about 58% of their time building new features, a share unchanged for nearly two years 1.

Molly Sands, who heads Atlassian's Teamwork Lab, names the mechanism 1. Time saved by automation goes wherever an organization's incentives point, and unchanged incentives send it into backlog processing rather than innovation 1. The flat allocation of engineering time is what that looks like in data 1.

Quality is decided upstream, where teams are weakest

Planning and documentation correlated more strongly with code quality than any other workplace dimension the index measured, yet only about 62% of organizations rate their own favorably 1. Incident handling scored highest: teams fight fires better than they specify what to build 1. Atlassian's CTO for AI and Teamwork, Taroon Mandhana, made the same point in September: the biggest bottleneck in AI software engineering "isn't model intelligence, it's organizational context" 3.

The seller of the bottleneck now publishes the benchmark

One conflict sits beside the data. The index runs on DX, an engineering-analytics business Atlassian bought, and every bottleneck it names is a category Atlassian sells into 14. Co-founder and chief executive Mike Cannon-Brookes introduced it personally 2, and on the November 5 earnings call he argued that a rise in agent-driven work increases the need to track and coordinate it 4.

More code brings a security bill and a pricing change

The flood of new code carries a security bill too: Atlassian's November bulletin covered fixes for 318 vulnerabilities, a reminder that more code, whatever wrote it, is more surface to defend 5. Monetization is changing anyway: usage-based pricing enforcement begins in early December, which Cannon-Brookes called "not a near-term revenue driver", with no meaningful fiscal 2027 revenue assumed 4.

The seat question is what February decides

The bear case on the stock says AI coding deletes Atlassian's per-seat subscriptions, one seat per displaced engineer. The index points the other way: engineering time redistributed rather than removed, and if that holds, seats hold too, with spending following the bottleneck into review and planning tools. Nobody has the answer yet; the fiscal Q2 report expected in February is where it gets tested 6.

The market has priced the fear first. The shares lost 34% in the month to November 17 and kept sliding the next day 67. They trade at 5.1 times trailing revenue, down from 6.5 at the end of September 6. Underneath, management guided cloud growth down from 31% to 27% for the December quarter 6.

Revenue growth held up while the shares fell by a third

Growth held up through the September quarter, before the shares fell a third

10%15%20%25%30%35%Q3 '24Q1 '25Q3 '25Q1 '26Q3 '2623.7%reported 5 Nov; the shares thenfell 34%
Data
Revenue growth, year over year
Q2 '2420.5%
Q3 '2421.5%
Q4 '2421.4%
Q1 '2514.1%
Q2 '2522.3%
Q3 '2520.6%
Q4 '2523.3%
Q1 '2631.7%
Q2 '2627.6%
Q3 '2623.7%
Total revenue, year-over-year change, percent, by calendar quarter; the last point is the quarter ended September 30, 2026. Sources: Atlassian SEC filings; share-price move per Trefis via Yahoo Finance.8,6

Operating margin flipped positive in the June quarter, before the selloff

Two quarters of double-digit operating margin, then a 34% slide in the shares

-10%-5%0%5%10%15%Q3 '24Q1 '25Q3 '25Q1 '26Q3 '2610.7%first double-digit operatingmargin
Data
Operating margin
Q2 '24-5.9%
Q3 '24-2.7%
Q4 '24-4.5%
Q1 '25-0.9%
Q2 '25-2.1%
Q3 '25-6.7%
Q4 '25-3%
Q1 '26-3.1%
Q2 '2611.9%
Q3 '2610.7%
Operating income as a percent of revenue, by calendar quarter; the last point is the quarter ended September 30, 2026. Source: Atlassian SEC filings.8

Measure what survives, not what was written

The index's method is the one worth copying: measure survivorship, not raw output 1. The scarce skills in software are now specifying, reviewing and documenting, and unlike model capability they can be learned, the one link in this chain a person still controls. Atlassian reports fiscal Q2 in February, the first quarter to include the new pricing, and the first real read on whether the seats held 6.

Get Platforms by email, free

The cloud, data and developer software AI is built on. In our daily email, every weekday.

Loading TEAM

Reader comments

0 comments

    More in Platforms

    All in this topic →

    Platforms

    Cloudflare's CFO says expect a 3% beat, not 6%, as machine traffic hits 68%

    The stock rallied toward the street's $350 median target hours after the warning, inside a company whose insiders sold roughly $330 million of stock in six months without a single purchase. The risk of a smaller beat sits with holders, not with the AI labs that will end up paying for the web.

    Platforms

    Replit's CEO says revenue went from $3 million to $1 billion in two years, 9 times sales vs Cursor's 30

    The first top line Replit has ever put on the record makes it the cheapest of the big AI app builders per dollar of revenue: roughly nine times sales, against about thirty for Cursor and Lovable. The gap runs through token prices, and Microsoft, now Replit's distributor and model rival, reset those last week.

    Platforms

    Exa says its agents wrote kernels that cut indexing cost 47% on Amazon's chips, not Nvidia's

    Exa's first public number on the cost of running its index was earned on Amazon's AI chips rather than Nvidia's, and it lands two weeks after DeepSeek, the Chinese AI lab, set a 75-cent floor under the price of search.

    Platforms

    Atlassian ships Secoda's data catalog for its AI agents; the test is holding ARR growth above 18.5%

    Ten months after quietly buying the data-catalog startup Secoda, Atlassian has wired data warehouses into the graph its agents read, at a price no filing discloses. The bet: whoever supplies agents their context gets paid whichever model wins, and investors who cut the stock 17 percent last week want the revenue proof.

    Platforms

    Dynatrace tells its largest customers to cut log costs in half while log consumption nears $250 million

    The advice comes from the vendor's own market chief, days after its CFO built an investor case on a log-consumption boom nearing $250 million. Until enterprises pay for the agent trace instead of the stored byte, the gap between those two stories sits with shareholders.

    Platforms

    Appian's first agent economics: an 80% token cut from giving agents less work

    The November 17 launch prices nothing, and the one named customer saved money by moving work off agents. That points demand at Appian's legacy process line, and APPN closed down 3.8%.

    Subscribe to TheInference for free