GitLab Inc. GTLB

Key data

The business

GitLab sells one thing: a single web application that a company's software teams live inside all day. They plan work, write and review code, run automated security scans, and ship the result, all in the same place. It is sold as an annual subscription priced per developer seat, in two paid tiers, Premium and Ultimate, sitting on top of a free version that seeds the funnel. The customers are enterprise engineering organizations, and the money is made by landing one team and then expanding across the company as more developers and more of the pipeline move onto the platform. The moat is switching cost: once GitLab is the system of record for how software gets built, pulling it out means retraining every engineer and rewiring every automation, so incumbency compounds quietly. The number that matters is not new logos, it is how much more each existing customer spends a year later.

The numbers

Revenue has climbed every quarter without a stumble, and the newest one is the largest yet. What changed this period is below the top line.

Revenue rose 21% year over year, yet the loss widened back to $37M from $9M, because operating expenses grew 32% while revenue grew 21%. The company had been walking its GAAP loss toward zero for six quarters, and Q2 FY27 is the quarter it walked back. That is the inflection to name out loud: not the revenue, which is steady, but the spending, which just outran it.

Revenue compounded about 31% a year over the last three fiscal years while the share count grew about 12%, and free cash flow per share flipped from −$0.56 to +$1.34. That is the whole re-rating story: the market decided GitLab turned into a cash machine.

Here is what the cash line does not say. That 20% cash margin is almost exactly matched by stock handed to employees, running near 22% of revenue. The margin the company reports and the margin the owner keeps are two different numbers, and after the stock goes out the door the owner keeps close to nothing. What this memo believes that the tape does not: the profitability turn is thinner than the free cash flow line advertises, and Q2's widening loss is the first crack in it. The print that settles it is the operating margin over the next two quarters against the direction of stock compensation.

Management

Insiders are heavy net sellers: 40 sales worth $63M against 4 buys worth $376K over the last year, a net drain of about 0.75% of the company, led by a board-linked holder trimming the same position again and again. The data pulled does not split the planned pre-scheduled sales from the discretionary ones, and that split matters, so read it as directional, not damning. More telling is the churn in the corner offices: the finance chair changed hands three times in a year and the technology chief was replaced, which is a lot of instability for a company this size. The buyback authorized last winter still sits untouched, zero shares repurchased this year, so the one shareholder-friendly lever announced has not been pulled while the share count keeps climbing.

How it fails or surprises you

The spending stays loose. Q2 FY27 grew revenue 21% and grew opex 32%, tripling the operating loss to $57M in a quarter that set a revenue record. If that is durable reinvestment, fine. If it is the cost of adding each new dollar creeping up, the exact way a software growth engine wears out, the cash margin gives it all back. The print is operating margin over the next two quarters.

Seats meet AI, and Microsoft. GitLab bills per developer. If AI coding tools mean fewer developers per customer, or GitHub, owned by Microsoft and bundled with Copilot, keeps undercutting on price, expansion inside existing accounts stalls. It shows first in current remaining performance obligations, not in revenue, which lags a quarter or two. Watch current RPO growth against revenue growth.

The AI add-on lands (right tail). GitLab is a rare independent, cash-generative platform for the whole software pipeline, and its Duo AI layer could put usage-based revenue on top of the seat, re-accelerating a decelerating line, while making it an obvious acquisition target. The market pays for neither today. The print is disclosed AI-attributed revenue or a reversal in the expansion rate.

Closing thoughts

At 8 times sales the tape prices continued low-20s% growth and a cash turn that holds. The growth is real and decelerating gently. The cash is real but hollowed out by stock, so the owner's true take is near zero, and the newest quarter went the wrong way on spending. Weigh that against the upside: a genuine AI product on a locked-in installed base, and scarcity value to a larger buyer. My read, and it is judgment not arithmetic, is that this is closer to fairly priced than cheap, with the left tail slightly the fatter one, because the reinvestment discipline that the cash story assumes just cracked.

The bet is still that big companies keep doing all their software work inside one GitLab subscription and keep paying for every developer who logs in. What breaks it is expansion inside those accounts slowing, and the pair that tells you first is current RPO growth against revenue growth, read alongside the operating margin quarter to quarter. If the next two quarters hold revenue in the low 20s% but the operating loss keeps widening and stock compensation stays near a fifth of sales, the cash-machine story was a mirage.