GitLab
The workshop where a company's software gets made: one application that stores the code, tests every change, checks it for security holes and ships it.
Our viewUnattractive Medium conviction
The scores
The profile · 1-4
Growth3/4 ↓
Growing 21%, but decelerating to about 15% on its own guide and narrowing to the hosted third of revenue.
Business quality3/4 ↓
The rival's own migration tool proves the exit cost, but retention has fallen two years running and the attacker now writes the code.
Financial position4/4 →
Net cash near a fifth of enterprise value, no borrowings of any kind, no maturities, and the plan needs no outside money.
Execution record3/4 →
Beat its own guidance eight quarters running, but the consumption metric was redefined in its second quarter and restated.
Per share capture2/4 →
Stock compensation is a quarter of revenue and twice reported profit; the buyback spends most of the cash flow to stand still.
Potential2/4 ↑
The seat pond is already smaller than the company, so the upside needs a consumption business that is mostly commitment today.
Risk profile3/4 →
No customer above a tenth of receivables, but most revenue is booked in one country and leans on public-sector budgets.
Valuation2/4 ↓
The price pays for a margin never earned on full costs, through a credits business that is mostly commitment rather than usage.
Evidence quality3/4 →
Every load-bearing number is filed, and several reach outside the company, but the AI metric the case turns on exists only on a call.
The Verdict
We rate GitLab unattractive. It sells the one application where companies store their code, test every change and ship it, and sales are growing faster than its own forecast, so the business is solid. The shares doubled in four months on the hope that AI coding agents will push far more work through it. But the AI business is still small, much of it money customers moved from existing licences, and the company forecasts lower profit this year. The price is full, not cheap: it assumes a profit level GitLab has never earned, so falling short would hurt. The view hinges on AI revenue proving to be new money.
The long caseMachines now write more code, pushing more changes through code platforms; but the AI money goes to model makers, so GitLab must show its AI revenue is new money.
The evidence under this - 7 claims
C01facthigh confidence
GitLab sells one application covering source control, code review, continuous integration, security scanning and deployment, priced per user seat.
C15facthigh confidence
Q2 FY2027 revenue was $286.3M, up 21.3% year on year, and about five points ahead of the company's own guidance.
C17facthigh confidence
FY2027 non-GAAP operating income is guided to $148M-$152M against $162.8M actually earned in FY2026, a fall of about 8% on revenue up 18% to 19%.
sourceGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The company told investors in May 2026 that it would reinvest the majority of restructuring savings, so the decline is a disclosed choice rather than a deterioration.
- The guide has been raised twice: from $129M-$137M in March to $135M-$141M in June to $148M-$152M now, so the direction of revision is up even though the level is below last year.
C18estimatehigh confidence
The same guidance implies second-half non-GAAP operating income of about $69.9M against $97.1M a year earlier, a fall of 28%.
derivedFY2027 guidance midpoint of $150M less first-half non-GAAP operating income of $80.1M from the Q2 release reconciliation. The prior-year half is FY2026's $162.765M from the July investor update less first-half FY2026 of $65.694M from the same reconciliation. Margin falls from 19.2% to about 12.0%.
against this
- The restructuring's remaining charges are excluded from the non-GAAP measure, so this is not a one-off cost effect but planned operating spend.
- Management has beaten its own non-GAAP operating income guidance in each of the last eight quarters, so the second-half figure may well be conservative.
C26management claimmedium confidence
Paid consumption run rate ended Q2 FY2027 above $40M, up from about $15M exiting Q1, against a stated year-end objective above $100M.
sourceGitLab Q2 FY2027 earnings call transcript, 1 September 2026primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The metric is not a filed figure and its definition was changed in its second quarter of life, to include Flex and exclude one-time incentives, with the Q1 figure restated to about $15M.
- The filed proxy for it is the $40.4M of non-cancellable Flex commitments in remaining performance obligations, of which only $15.0M had been billed - so most of the run rate is a commitment rather than consumption.
C27inferencemedium confidence
Most of the jump in paid consumption run rate is seat budget redirected into Flex, not new spending: management says customers may fund Flex with their previous seat subscription dollars.
derivedPaid consumption run rate rose from about $15M to more than $40M and management said the rise was 'thanks to the introduction of Flex'. Flex commitments were $40.4M in remaining performance obligations at 31 July 2026. The chief executive said customers 'may start to fund their flex agreement with their previous seat based subscription dollars', and the chief financial officer guided to 'limited contribution' from Duo Agent Platform in FY2027.
against this
- Duo Agent Platform's own paid consumption run rate grew about 50% sequentially, which is a credits number that predates Flex and cannot be redirected seat spend.
- One top-20 US commercial bank expanded its credit pool nearly tenfold in the quarter, which is new money by any reading, though it is a single account.
C35estimatehigh confidence
GitLab's enterprise value is about $6.68B, roughly 5.9 times guided FY2027 revenue and 44.5 times guided operating profit before stock costs, after a 107.3% gain in 120 sessions.
derivedThe 2026-09-08 close of $47.42 from the pack on the 167,281 thousand shares outstanding in the 10-Q gives $7.93B of market value; less $1.257B of cash and investments and no debt gives $6.68B. Guidance midpoints are $1,131M of revenue and $150M of non-GAAP operating income. The 120-session return is from the pack's market tier.
against this
- The 252-session return is only 3.7%, so the doubling is a recovery from a low of $18.73 rather than a fresh advance, and the shares are still 14.6% below their 52-week high.
- On revenue GitLab is the cheaper of the two comparable companies its own annual report names, and it trades at well under half its own six-year median multiple of sales.
The company
Making software at a company is a chain of steps: store the code, review each change, build and test it, scan it for security holes, then ship it. GitLab sells one application that does all of them under one set of permissions and one audit record, priced per user.
The alternative is a stitched-together set of separate tools, or tools the company builds itself. Two thirds of GitLab's revenue comes from customers who run it on their own computers.
The evidence under this - 3 claims
C01facthigh confidence
GitLab sells one application covering source control, code review, continuous integration, security scanning and deployment, priced per user seat.
C02facthigh confidence
Its buyers otherwise stitch together separate tools, or build their own; JFrog's 10-K says home-grown alternatives often do not scale.
C03facthigh confidence
Two thirds of GitLab's Q2 FY2027 revenue is software customers run on their own infrastructure: $181.4M of the $286.3M total.
How it makes money
Customers pay per user each year, in three tiers. The big ones are growing fastest: 1,571 customers now pay more than $100,000 a year, up 16.9%. Since the second quarter customers can also commit a dollar sum, called Flex, and draw it down as their people and AI agents use the tools.
The evidence under this - 3 claims
C01facthigh confidence
GitLab sells one application covering source control, code review, continuous integration, security scanning and deployment, priced per user seat.
C30facthigh confidence
Customers paying more than $100,000 of ARR reached 1,571, up 16.9%, while those paying more than $5,000 reached 11,114, up 8%.
C13facthigh confidence
GitLab's remaining performance obligations were about $1.2B at 31 July 2026, of which $40.4M is non-cancellable Flex commitments and only $15.0M of that had been billed.
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 FY2027) | How it earns it |
|---|---|---|
| Self-managed subscriptions | 56.1% | Yearly per-seat subscriptions for GitLab installed on the customer's own servers or cloud, excluding the licence portion booked upfront. |
| SaaS | 34.1% | Per-seat subscriptions to GitLab hosted by GitLab, shared or single-tenant (Dedicated), recognised over the term. |
| Self-managed licences | 7.2% | The licence element of a self-managed subscription, recognised upfront when the software is delivered. |
| Professional services and other | 2.5% | Training, migration and implementation work sold alongside the platform. |
Who runs it
William Staples has run GitLab since December 2024, after running New Relic; Jessica Ross became finance chief in January 2026. Co-founder Sytse Sijbrandij, now executive chair, controls just over half the votes through shares carrying ten votes each, so outside shareholders cannot outvote him. Most of the bonus rides on new recurring sales, the rest on adjusted profit; none on cash or on how many new shares staff receive. Last year's performance shares were forfeited on a missed sales goal.
The evidence under this - 3 claims
C45facthigh confidence
William Staples has been chief executive since December 2024, after running New Relic from July 2021 to December 2023; Jessica Ross has been chief financial officer since 15 January 2026, from Frontdoor, and also became principal accounting officer when the chief accounting officer left on 28 August 2026; co-founder Sytse Sijbrandij, chief executive until December 2024, is executive chair.
C43facthigh confidence
At 1 April 2026 the 12 directors and officers as a group held 1,002,259 Class A and 17,072,653 Class B shares, 54.14% of the voting power; co-founder Sytse Sijbrandij held 16,750,651 Class B shares including 1.5M options, 50.43% of the votes and about a tenth of the shares; Class B carries ten votes a share.
C44facthigh confidence
GitLab's FY2026 annual bonus paid on Net ARR (70%) and non-GAAP operating income margin (30%) and paid 74.51% of target; long-term awards were 60% RSUs and 40% PSUs earned on one-year Net ARR, capped at target if a non-GAAP operating income threshold is missed, and the FY2026 PSUs, including a third of the CEO's new-hire award, were forfeited below the Net ARR threshold.
against this
- No metric pays on GAAP profit, free cash flow, stock compensation or the share count, so the dilution C21 records costs the plan nothing.
- The Net ARR targets are not disclosed, so whether the forfeiture reflects a missed plan or a stretch goal cannot be checked.
Why it wins, and who could beat it
Why customers come back
Code moves between platforms in an afternoon. What does not move is everything around it. GitHub's own migration tool, built by the rival with every reason to make leaving look easy, leaves behind the pipelines, security history, approval rules and settings, which a regulated buyer must rebuild by hand.
So customers stay: more than 90% of revenue renews, and existing customers spend 17% more than a year before. Two years ago that was 23%.
The evidence under this - 3 claims
C04facthigh confidence
GitHub's own migration tool does not carry over CI/CD pipelines, vulnerability reports, repository policies, group settings, CI/CD variables or job artifacts.
C05facthigh confidence
Gross retention stayed above 90% in Q2 FY2027 and dollar-based net retention was 117%.
sourceGitLab Form 10-Q, quarter ended 31 July 2026primaryGitLab Q2 FY2027 earnings call transcript, 1 September 2026primary
against this
- Dollar-based net retention was 121% a year earlier and 123% at the end of FY2025, so the same metric has fallen in each of the last two years even as it rose sequentially this quarter.
- The count of customers paying more than $5,000 of ARR grew only 8% year on year, so expansion is concentrating in existing large accounts rather than broadening.
C29facthigh confidence
Dollar-based net retention has fallen in each of the last two years: 123% for FY2025, 118% for FY2026, and 117% at 31 July 2026 against 121% a year earlier.
The threat
The danger is that the work moves while the code stays. GitHub's coding agent plans a change, writes it and tests it inside GitHub, and opened more than a million change requests in four months of 2025. Inside GitLab, about 80% of use of its new AI context product comes from customers plugging in rival agents such as Claude Code.
The evidence under this - 3 claims
C06facthigh confidence
GitHub's Copilot coding agent researches a repository, plans a change and commits it on a branch, running on GitHub's own CI runners.
against this
- The same documentation caps a run at 59 minutes in an ephemeral environment, so it is an assistant inside one repository rather than a replacement for a governed pipeline across an estate.
- GitLab's own answer is that it is model and cloud neutral and can be deployed air-gapped, which GitHub's hosted agent is not.
C07factmedium confidence
More than one million pull requests on GitHub were created by its coding agent between May and September 2025, on a base of over 180 million developers.
C08management claimmedium confidence
About 80% of query volume on GitLab Orbit, its new context product, comes from customers connecting rival agents such as Claude Code.
against this
- Management presents this as evidence of neutrality and a reason to buy credits, and it is equally evidence that the code-writing job has moved to somebody else's product while GitLab keeps the storage and governance layer.
- Orbit was in public beta all quarter, so the query mix is drawn from early adopters and need not hold as it becomes generally available.
The rivals: MSFT, FROG, TEAM, and 1 more that is private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | How far along | Priced against it |
|---|---|---|---|
| MSFT | GitHub Copilot coding agent (Microsoft)Priced against it: Shown, not counted | On sale | Shown, not counted |
| Private | Anthropic Claude Code and other external coding agentsPriced against it: Private | On sale | Private |
| FROG | JFrog ArtifactoryPriced against it: Yes | On sale | Yes |
| TEAM | AtlassianPriced against it: Yes | On sale | Yes |
How big it could get
How big this could get
Price every American software developer, 1,905,400 of them, at GitHub's entry enterprise price of $252 a year and the pool comes to about $480M. GitLab already earned $955.2M last year, twice that, so American seats are not the ceiling. The wider bound is global: GitHub counts more than 180 million developer accounts, and added 36 million in a year.
The evidence under this - 5 claims
C10facthigh confidence
The US Bureau of Labor Statistics counted 1,905,400 software developers, quality assurance analysts and testers employed in 2025, and projects 10% growth to 2035.
C11factmedium confidence
GitHub's published list price for its Enterprise plan starts at $21 per user per month, or $252 a year.
C12estimatemedium confidence
GitLab's FY2026 revenue of $955.2M is about twice the $480M that every employed US software developer would be worth at the nearest rival's entry enterprise seat price.
derivedC10 multiplied by C11 gives 1,905,400 x $252 = $480.2M of annual US seat spend at that price point. GitLab's FY2026 revenue is the filed figure in the 10-K. The ratio is 2.0x.
against this
- GitHub's entry price is a floor, not the market clearing price: GitLab's own tiers sell well above it and the ratio would invert at a higher price point, which is exactly why the prize does not compose at any single price.
- GitLab sells outside the United States - 16.4% of Q2 FY2027 revenue was booked through European and Asia-Pacific entities - so the denominator understates its addressable seat base.
C40facthigh confidence
GitLab's revenue for the fiscal year ended 31 January 2026 was $955.2M.
C14factmedium confidence
More than 36 million new developers joined GitHub in the year to Octoverse 2025, against a base of more than 180 million.
Who has signed
Customers have committed about $1.2B not yet counted as revenue, mostly ordinary seat subscriptions. The AI part is Flex: $40.4M of commitments, of which only $15.0M had been billed by 31 July. Management wants paid use above $100M a year by January, against more than $40M now.
The evidence under this - 2 claims
C13facthigh confidence
GitLab's remaining performance obligations were about $1.2B at 31 July 2026, of which $40.4M is non-cancellable Flex commitments and only $15.0M of that had been billed.
C26management claimmedium confidence
Paid consumption run rate ended Q2 FY2027 above $40M, up from about $15M exiting Q1, against a stated year-end objective above $100M.
sourceGitLab Q2 FY2027 earnings call transcript, 1 September 2026primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The metric is not a filed figure and its definition was changed in its second quarter of life, to include Flex and exclude one-time incentives, with the Q1 figure restated to about $15M.
- The filed proxy for it is the $40.4M of non-cancellable Flex commitments in remaining performance obligations, of which only $15.0M had been billed - so most of the run rate is a commitment rather than consumption.
The results
The numbers
Is it growing?Steady
Sales grew 23% over the twelve months to Jul 2026, about the same pace as the 26% in the year to Jan 2026.
Is it profitable?Losing
Operating margin is −8.5%, up from −32% in the year to Jan 2024.
Does growth pay off?Pays off
Running costs take 94¢ of each sales dollar, down from 139¢ in the year to Jan 2022.
Does it earn good returns?Poor
Shareholders' money lost 5.9% a year, up from −57% in the year to Jan 2022.
Is it making cash?Strong
It turned 20% of sales into free cash, $214M over the last twelve months.
How steady is it?Swingy
Yearly sales growth has ranged from +26% to +68% over the last 5 years.
Are shares being diluted?Diluting
The share count rose 112% in 4.5 years, and pay in shares costs 22% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in January.
| Jan 2022 | Jan 2023 | Jan 2024 | Jan 2025 | Jan 2026 | 12 mo to Jul 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $253M | $424M | $580M | $759M | $955M | $1.06B |
| Sales growth | +66% | +68% | +37% | +31% | +26% | +23% |
| Profit | ||||||
| Gross margin | 88% | 88% | 90% | 89% | 87% | 86% |
| Operating margin | −51% | −50% | −32% | −19% | −7.4% | −8.5% |
| Net margin | −61% | −41% | −73% | −0.8% | −5.9% | −5.0% |
| What growth costs, per sales dollar | ||||||
| Research and development | 38% | 37% | 35% | 32% | 29% | 29% |
| Sales and marketing | 76% | 73% | 61% | 51% | 46% | 45% |
| General and admin | 25% | 28% | 26% | 25% | 20% | 21% |
| Pay in shares | 12% | 29% | 28% | 24% | 23% | 22% |
| Capital spending | 1.4% | 1.4% | 0.3% | 0.5% | 1.1% | 0.9% |
| Cash | ||||||
| Cash from operations | −$49.8M | −$77.4M | $35.0M | −$64.0M | $233M | $223M |
| Free cash flow | −$53.4M | −$83.5M | $33.4M | −$67.7M | $222M | $214M |
| Balance sheet | ||||||
| Cash and short-term investments | $935M | $295M | $288M | $228M | $230M | $226M |
| Returns and shares | ||||||
| Shares (diluted) | 79.8M | 148M | 154M | 161M | 167M | 169M |
What growth costs
Each new sale has cost GitLab less to win: selling, research, admin and stock pay all shrank as a share of sales over three years, so losses narrowed. The cost of delivering the product now grows faster than sales, and this year's plan spends the savings rather than keeping them.
The evidence under this - 3 claims
C42facthigh confidence
Per sales dollar, sales and marketing fell from 73.1 cents in FY2023 to 44.7 over the four quarters to July 2026, R&D from 36.8 to 28.8, general and administrative from 27.8 to 20.8 and stock pay from 28.9 to 21.8, while cost of revenue rose from 12.2 to 14.2 and operating margin improved from -49.8% to -8.5%.
sourceGitLab Form 10-K, fiscal year ended 31 January 2026primaryGitLab Form 10-K, fiscal year ended 31 January 2025primaryGitLab Form 10-Q, quarter ended 31 July 2026primary
against this
- Cost of revenue is growing faster than sales, which is the gross-margin fall C22 records, so the leverage is all below gross profit.
- The FY2027 guide takes non-GAAP operating income down on revenue up 18% to 19% (C17), so the trend is being reversed this year by choice.
C22facthigh confidence
GAAP gross margin fell to 84% from 88% and the non-GAAP figure to 86.5% from 90%, as the cost of subscription revenue rose 75.9% on subscription revenue up 21.5%.
sourceGitLab Form 10-Q, quarter ended 31 July 2026primaryGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab Q2 FY2027 earnings call transcript, 1 September 2026primary
against this
- The chief financial officer attributed the move to mix, saying SaaS is now 34% of revenue against 22% at the initial public offering and 'we expect those margins to come down over time'.
- The chief executive added that for many customers the inference cost is not embedded in the GitLab agreement at all, since they bring their own models, which would keep agent workloads off GitLab's cost line.
C17facthigh confidence
FY2027 non-GAAP operating income is guided to $148M-$152M against $162.8M actually earned in FY2026, a fall of about 8% on revenue up 18% to 19%.
sourceGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The company told investors in May 2026 that it would reinvest the majority of restructuring savings, so the decline is a disclosed choice rather than a deterioration.
- The guide has been raised twice: from $129M-$137M in March to $135M-$141M in June to $148M-$152M now, so the direction of revision is up even though the level is below last year.
The last report (1 Sep)
Quarter ended July 2026, filed figures.
21%
Sales growth
against the same quarter a year earlier
8.4%
Growth from the quarter before
against 1.4% the quarter before that
-20¢
Operating profit
per dollar of sales
In the second quarter of fiscal 2027, to 31 July, sales grew 21.3% to $286.3M and the shares jumped about 21%. The forecast told a different story: second-half growth slows to about 15%, and profit before stock pay falls 28% from a year earlier.
The cost of serving customers rose 75.9% against subscription sales up 21.5%, so the share of each sales dollar left after those costs fell from 90 cents to 86.5.
What it did to the thesisIt supported the case against the shares. Sales beat, but the company's own forecast has profit falling in the second half, and serving customers is getting more expensive as the newer services grow. The price assumes profit the business has never earned, which keeps the view Unattractive.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $0.24 against $0.18 expected; $-0.22 as filed, the gap being what the company adds back.
The evidence under this - 5 claims
C15facthigh confidence
Q2 FY2027 revenue was $286.3M, up 21.3% year on year, and about five points ahead of the company's own guidance.
C16estimatehigh confidence
FY2027 revenue guidance of $1,129M-$1,133M requires second-half revenue of $578.6M-$582.6M against $504.8M, growth of 14.6% to 15.4%.
derivedGuidance from the Q2 FY2027 release, less first-half revenue of $550.412M from the 10-Q. The prior-year comparison is Q3 FY2026 of $244.353M plus Q4 FY2026 of $260.402M. First-half growth was 22.2%, so the guide implies about seven points of deceleration.
against this
- Management said explicitly that the second-half outlook assumes a more normalised pace of bookings rather than a repeat of Q2, so the deceleration is deliberate conservatism rather than observed weakness.
- The guide excludes any Flex revenue-recognition impact, which management sizes at up to $13M and which would run the other way if adoption is slower than they now expect.
C18estimatehigh confidence
The same guidance implies second-half non-GAAP operating income of about $69.9M against $97.1M a year earlier, a fall of 28%.
derivedFY2027 guidance midpoint of $150M less first-half non-GAAP operating income of $80.1M from the Q2 release reconciliation. The prior-year half is FY2026's $162.765M from the July investor update less first-half FY2026 of $65.694M from the same reconciliation. Margin falls from 19.2% to about 12.0%.
against this
- The restructuring's remaining charges are excluded from the non-GAAP measure, so this is not a one-off cost effect but planned operating spend.
- Management has beaten its own non-GAAP operating income guidance in each of the last eight quarters, so the second-half figure may well be conservative.
C22facthigh confidence
GAAP gross margin fell to 84% from 88% and the non-GAAP figure to 86.5% from 90%, as the cost of subscription revenue rose 75.9% on subscription revenue up 21.5%.
sourceGitLab Form 10-Q, quarter ended 31 July 2026primaryGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab Q2 FY2027 earnings call transcript, 1 September 2026primary
against this
- The chief financial officer attributed the move to mix, saying SaaS is now 34% of revenue against 22% at the initial public offering and 'we expect those margins to come down over time'.
- The chief executive added that for many customers the inference cost is not embedded in the GitLab agreement at all, since they bring their own models, which would keep agent workloads off GitLab's cost line.
C23estimatemedium confidence
The incremental subscription dollar carried about 70 cents of gross profit against 85 cents on the base, once restructuring and stock costs are removed.
derivedSubscription revenue rose $45.627M year on year and its cost rose $16.505M. Removing $1.023M of restructuring and the $1.630M increase in stock compensation charged to cost of revenue leaves $13.852M of cash cost on $45.627M of new revenue, an incremental gross margin of 69.6%. Subscription gross margin on the whole base was 85.2%.
against this
- A quarter is a short window for an incremental margin and hosting costs are lumpy: a single capacity commitment or a dedicated-instance ramp lands in one quarter and amortises over several.
- The 10-Q names both SaaS and Duo Agent Platform as drivers of cloud cost, so part of the gap is the mix shift management describes rather than agent inference.
What management did with the money
It holds $1.26B of cash and no debt. In the first half it spent $154.7M buying back shares while issuing $125.1M to staff, shrinking the count 1.6%. It also cut about 14% of its workforce and plans to reinvest most of the savings.
The evidence under this - 3 claims
C31facthigh confidence
GitLab held $1.26B of cash and short-term investments at 31 July 2026 with no borrowings on its balance sheet.
C32facthigh confidence
GitLab spent $154.7M on buybacks in the first half while issuing $125.1M of stock to employees, shrinking the share count 1.6%.
C34management claimhigh confidence
GitLab cut about 14% of its workforce, 350 people, and is exiting 22 countries, for $30M-$35M of charges, intending to reinvest the majority of the savings.
sourceGitLab Q1 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab 'Act 2' letter to customers and investors, Form 8-K…primary
against this
- The reinvestment promise is testable and the FY2027 operating income guide is consistent with it, so the claim has already been partly kept rather than merely stated.
- Retention incentives run to the first quarter of FY2028 and the plan is only 'substantially complete' by the end of FY2027, so the cost is not finished and the savings are not yet visible in a full quarter.
The price, and where we disagree
In short. Full. The price pays for a level of profit GitLab has never earned, relying on an AI business it has barely started billing for.
Share price
$47.42
close,
Company value for each $1 of sales
7.4
rivals' middle 12.4
The shares more than doubled in 120 trading days, pricing the AI story before the AI revenue arrived. Analysts have not followed: 11 rate them a buy, 18 a hold, 1 a sell.
The first test is the third-quarter report, due by 1 December: is paid AI use growing on new budgets, or on seat money moved across?
The evidence under this - 4 claims
C35estimatehigh confidence
GitLab's enterprise value is about $6.68B, roughly 5.9 times guided FY2027 revenue and 44.5 times guided operating profit before stock costs, after a 107.3% gain in 120 sessions.
derivedThe 2026-09-08 close of $47.42 from the pack on the 167,281 thousand shares outstanding in the 10-Q gives $7.93B of market value; less $1.257B of cash and investments and no debt gives $6.68B. Guidance midpoints are $1,131M of revenue and $150M of non-GAAP operating income. The 120-session return is from the pack's market tier.
against this
- The 252-session return is only 3.7%, so the doubling is a recovery from a low of $18.73 rather than a fresh advance, and the shares are still 14.6% below their 52-week high.
- On revenue GitLab is the cheaper of the two comparable companies its own annual report names, and it trades at well under half its own six-year median multiple of sales.
C36factmedium confidence
The sell-side panel the pack aggregates sits at Hold - 11 buy, 18 hold, 1 sell - and short interest was 11.0% of the free float at the 14 August 2026 settlement.
C26management claimmedium confidence
Paid consumption run rate ended Q2 FY2027 above $40M, up from about $15M exiting Q1, against a stated year-end objective above $100M.
sourceGitLab Q2 FY2027 earnings call transcript, 1 September 2026primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The metric is not a filed figure and its definition was changed in its second quarter of life, to include Flex and exclude one-time incentives, with the Q1 figure restated to about $15M.
- The filed proxy for it is the $40.4M of non-cancellable Flex commitments in remaining performance obligations, of which only $15.0M had been billed - so most of the run rate is a commitment rather than consumption.
C27inferencemedium confidence
Most of the jump in paid consumption run rate is seat budget redirected into Flex, not new spending: management says customers may fund Flex with their previous seat subscription dollars.
derivedPaid consumption run rate rose from about $15M to more than $40M and management said the rise was 'thanks to the introduction of Flex'. Flex commitments were $40.4M in remaining performance obligations at 31 July 2026. The chief executive said customers 'may start to fund their flex agreement with their previous seat based subscription dollars', and the chief financial officer guided to 'limited contribution' from Duo Agent Platform in FY2027.
against this
- Duo Agent Platform's own paid consumption run rate grew about 50% sequentially, which is a credits number that predates Flex and cannot be redirected seat spend.
- One top-20 US commercial bank expanded its credit pool nearly tenfold in the quarter, which is new money by any reading, though it is a single account.
How we read the multiple
Counting cash, a buyer pays about $7 for each dollar of last year's sales, and about $45 for each dollar of this year's expected profit before the cost of paying staff in shares.
Count that share pay as the cost it is and GitLab loses about $150M a year from operations, and its forecast says the second half gets worse.
So the price needs a profit that has never existed, from an AI business billing $40M a year.
Against its own history. Far cheaper than its own past on sales, but that past priced a faster-growing company. A lower price for slower growth is not a bargain.
The evidence under this - 4 claims
C35estimatehigh confidence
GitLab's enterprise value is about $6.68B, roughly 5.9 times guided FY2027 revenue and 44.5 times guided operating profit before stock costs, after a 107.3% gain in 120 sessions.
derivedThe 2026-09-08 close of $47.42 from the pack on the 167,281 thousand shares outstanding in the 10-Q gives $7.93B of market value; less $1.257B of cash and investments and no debt gives $6.68B. Guidance midpoints are $1,131M of revenue and $150M of non-GAAP operating income. The 120-session return is from the pack's market tier.
against this
- The 252-session return is only 3.7%, so the doubling is a recovery from a low of $18.73 rather than a fresh advance, and the shares are still 14.6% below their 52-week high.
- On revenue GitLab is the cheaper of the two comparable companies its own annual report names, and it trades at well under half its own six-year median multiple of sales.
C21facthigh confidence
Stock compensation was $75.0M in Q2 FY2027, 26.2% of revenue, up 38% year on year against revenue up 21%.
C18estimatehigh confidence
The same guidance implies second-half non-GAAP operating income of about $69.9M against $97.1M a year earlier, a fall of 28%.
derivedFY2027 guidance midpoint of $150M less first-half non-GAAP operating income of $80.1M from the Q2 release reconciliation. The prior-year half is FY2026's $162.765M from the July investor update less first-half FY2026 of $65.694M from the same reconciliation. Margin falls from 19.2% to about 12.0%.
against this
- The restructuring's remaining charges are excluded from the non-GAAP measure, so this is not a one-off cost effect but planned operating spend.
- Management has beaten its own non-GAAP operating income guidance in each of the last eight quarters, so the second-half figure may well be conservative.
C26management claimmedium confidence
Paid consumption run rate ended Q2 FY2027 above $40M, up from about $15M exiting Q1, against a stated year-end objective above $100M.
sourceGitLab Q2 FY2027 earnings call transcript, 1 September 2026primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The metric is not a filed figure and its definition was changed in its second quarter of life, to include Flex and exclude one-time incentives, with the Q1 figure restated to about $15M.
- The filed proxy for it is the $40.4M of non-cancellable Flex commitments in remaining performance obligations, of which only $15.0M had been billed - so most of the run rate is a commitment rather than consumption.
Priced against its rivals
The competitors its own annual report names, each confirmed against the rival's filing. Measured on company value for each $1 of sales, cheapest first:
| Ticker | Company | company value for each $1 of sales |
|---|---|---|
| TEAM | Atlassian | 7.2× |
| GTLB | GitLab | 7.4× |
| FROG | JFrog | 17.6× |
| MSFT | Microsoft not in the median | 11.3× |
The middle of the other 2 is 12.4. Microsoft Corporation competes with it in one product line inside a much bigger company, so its price says little about this business and it is not counted in the middle. GitLab Inc. is 2nd cheapest of 3.
Rivals' figures come from a data provider, so read the spread of the group rather than any single pair.
Level with Atlassian and well below JFrog on sales, but the slowest grower of the three, so the lower price is its own, and earned.
Private or foreign rivals that cannot be priced: Harness Inc., Anysphere (Cursor), Anthropic (Claude Code), OpenAI (Codex).
Where we disagree with the market
The market believes
AI is a tailwind: agents write more code, all of it lands in GitLab, so growth and profit speed up again.
We believe
The tailwind shows up in usage, not profit. This year the company's own preferred profit measure falls.
The raise was on sales. The same release guides second-half profit down 28%, a number that takes two subtractions to find and nobody printed. The new AI dollars are largely customers moving seat budgets into Flex, which management says they may do.
We are wrong if paid AI use passes $100M a year on new budgets, with second-half margins above the forecast.
What analysts say, and the evidence
The evidence under this - 4 claims
C17facthigh confidence
FY2027 non-GAAP operating income is guided to $148M-$152M against $162.8M actually earned in FY2026, a fall of about 8% on revenue up 18% to 19%.
sourceGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The company told investors in May 2026 that it would reinvest the majority of restructuring savings, so the decline is a disclosed choice rather than a deterioration.
- The guide has been raised twice: from $129M-$137M in March to $135M-$141M in June to $148M-$152M now, so the direction of revision is up even though the level is below last year.
C18estimatehigh confidence
The same guidance implies second-half non-GAAP operating income of about $69.9M against $97.1M a year earlier, a fall of 28%.
derivedFY2027 guidance midpoint of $150M less first-half non-GAAP operating income of $80.1M from the Q2 release reconciliation. The prior-year half is FY2026's $162.765M from the July investor update less first-half FY2026 of $65.694M from the same reconciliation. Margin falls from 19.2% to about 12.0%.
against this
- The restructuring's remaining charges are excluded from the non-GAAP measure, so this is not a one-off cost effect but planned operating spend.
- Management has beaten its own non-GAAP operating income guidance in each of the last eight quarters, so the second-half figure may well be conservative.
C27inferencemedium confidence
Most of the jump in paid consumption run rate is seat budget redirected into Flex, not new spending: management says customers may fund Flex with their previous seat subscription dollars.
derivedPaid consumption run rate rose from about $15M to more than $40M and management said the rise was 'thanks to the introduction of Flex'. Flex commitments were $40.4M in remaining performance obligations at 31 July 2026. The chief executive said customers 'may start to fund their flex agreement with their previous seat based subscription dollars', and the chief financial officer guided to 'limited contribution' from Duo Agent Platform in FY2027.
against this
- Duo Agent Platform's own paid consumption run rate grew about 50% sequentially, which is a credits number that predates Flex and cannot be redirected seat spend.
- One top-20 US commercial bank expanded its credit pool nearly tenfold in the quarter, which is new money by any reading, though it is a single account.
C26management claimmedium confidence
Paid consumption run rate ended Q2 FY2027 above $40M, up from about $15M exiting Q1, against a stated year-end objective above $100M.
sourceGitLab Q2 FY2027 earnings call transcript, 1 September 2026primaryGitLab July 2026 investor update, Form 8-K Exhibit 99.1primary
against this
- The metric is not a filed figure and its definition was changed in its second quarter of life, to include Flex and exclude one-time incentives, with the Q1 figure restated to about $15M.
- The filed proxy for it is the $40.4M of non-cancellable Flex commitments in remaining performance obligations, of which only $15.0M had been billed - so most of the run rate is a commitment rather than consumption.
What could go wrong, and how we would know
The risks
- AI raising the cost of serving. Each new subscription dollar keeps about 70 cents after serving costs, against 85 on the old ones.
- Customers adding less. Existing customers' extra spending has slowed for two years, from 23% to 17%.
- Stock pay outrunning sales. It grew 38% in a year against sales up 21%, and is now a quarter of revenue.
- A slower second half. The forecast has growth easing to about 15% in the second half, from 21.3% in the second quarter.
The evidence under this - 5 claims
C22facthigh confidence
GAAP gross margin fell to 84% from 88% and the non-GAAP figure to 86.5% from 90%, as the cost of subscription revenue rose 75.9% on subscription revenue up 21.5%.
sourceGitLab Form 10-Q, quarter ended 31 July 2026primaryGitLab Q2 FY2027 results release, Form 8-K Exhibit 99.1primaryGitLab Q2 FY2027 earnings call transcript, 1 September 2026primary
against this
- The chief financial officer attributed the move to mix, saying SaaS is now 34% of revenue against 22% at the initial public offering and 'we expect those margins to come down over time'.
- The chief executive added that for many customers the inference cost is not embedded in the GitLab agreement at all, since they bring their own models, which would keep agent workloads off GitLab's cost line.
C29facthigh confidence
Dollar-based net retention has fallen in each of the last two years: 123% for FY2025, 118% for FY2026, and 117% at 31 July 2026 against 121% a year earlier.
C21facthigh confidence
Stock compensation was $75.0M in Q2 FY2027, 26.2% of revenue, up 38% year on year against revenue up 21%.
C16estimatehigh confidence
FY2027 revenue guidance of $1,129M-$1,133M requires second-half revenue of $578.6M-$582.6M against $504.8M, growth of 14.6% to 15.4%.
derivedGuidance from the Q2 FY2027 release, less first-half revenue of $550.412M from the 10-Q. The prior-year comparison is Q3 FY2026 of $244.353M plus Q4 FY2026 of $260.402M. First-half growth was 22.2%, so the guide implies about seven points of deceleration.
against this
- Management said explicitly that the second-half outlook assumes a more normalised pace of bookings rather than a repeat of Q2, so the deceleration is deliberate conservatism rather than observed weakness.
- The guide excludes any Flex revenue-recognition impact, which management sizes at up to $13M and which would run the other way if adoption is slower than they now expect.
C15facthigh confidence
Q2 FY2027 revenue was $286.3M, up 21.3% year on year, and about five points ahead of the company's own guidance.
What we still do not know
- How much of the $40.4M of Flex commitments is money customers were not already spending on seats. Nothing filed separates it.
- How much paid work AI agents will leave inside GitLab, when customers buy the AI itself from the model makers.
- How many shares the company will count this year: its own assumption moved from 175 million to 166 million to 172 million in six months.
The evidence under this - 4 claims
C27inferencemedium confidence
Most of the jump in paid consumption run rate is seat budget redirected into Flex, not new spending: management says customers may fund Flex with their previous seat subscription dollars.
derivedPaid consumption run rate rose from about $15M to more than $40M and management said the rise was 'thanks to the introduction of Flex'. Flex commitments were $40.4M in remaining performance obligations at 31 July 2026. The chief executive said customers 'may start to fund their flex agreement with their previous seat based subscription dollars', and the chief financial officer guided to 'limited contribution' from Duo Agent Platform in FY2027.
against this
- Duo Agent Platform's own paid consumption run rate grew about 50% sequentially, which is a credits number that predates Flex and cannot be redirected seat spend.
- One top-20 US commercial bank expanded its credit pool nearly tenfold in the quarter, which is new money by any reading, though it is a single account.
C13facthigh confidence
GitLab's remaining performance obligations were about $1.2B at 31 July 2026, of which $40.4M is non-cancellable Flex commitments and only $15.0M of that had been billed.
C08management claimmedium confidence
About 80% of query volume on GitLab Orbit, its new context product, comes from customers connecting rival agents such as Claude Code.
against this
- Management presents this as evidence of neutrality and a reason to buy credits, and it is equally evidence that the code-writing job has moved to somebody else's product while GitLab keeps the storage and governance layer.
- Orbit was in public beta all quarter, so the query mix is drawn from early adopters and need not hold as it becomes generally available.
C33facthigh confidence
The diluted share count GitLab assumes for FY2027 has moved from 175 million in March to 166 million in June to 172 million in September.
What would change our view
- Unattractive→AttractiveConsumption becomes a business and the full-cost loss closes
It needs all of:
- ○Paid AI use ends the fiscal year, in January, above $70M a year, against roughly $40M today.checked at the next results, due by 1 Dec
- ○Stock pay falls below 25% of sales through fiscal 2028, so the gap to a real profit is closing.checked at the next results, due by 1 Dec
- ○The company's value falls to about 3.5 times this year's expected sales.about ≈$27.97 today, 41% below the last close
- Reopens the researchThe Flex commitments turn out to be new money
It needs:
- ○GitLab says how much of its new Flex commitments is fresh money rather than per-user budget moved across.checked at the next results, due by 1 Dec
What is coming
- 1 December 2026 Q3 FY2027 results: Paid consumption run rate against the $100M year-end objective and whether it is new budget or converted seats, and non-GAAP operating and gross margin against the guide.
Every dated event ahead
- 1 December 2026 Q3 FY2027 results: Paid consumption run rate against the $100M year-end objective and whether it is new budget or converted seats, and non-GAAP operating and gross margin against the guide.
- March 2027 Q4 FY2027 results and FY2028 guidance: Whether the $100M consumption objective was met, whether restructuring savings were reinvested or kept, and whether next year's guide resumes margin widening.
- May 2027 2027 proxy statement: Whether FY2027 Net ARR cleared the PSU threshold after the FY2026 forfeiture, and whether pay adds any per-share, cash or dilution measure.
Since we wrote this
Written
8 Sep
Price the thesis read
$47.42
8 Sep
- 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 8 SepThesis writtenUnattractive, medium convictionWritten a week after a quarter that beat on sales and lifted the shares about 21%, while the same forecast cut this year's profit.
Also checked 3 times with nothing new to act on.
Sources 21
Where this came from
Sources - 21
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-03-17Item 1 competition section naming Microsoft, Atlassian, JFrog and Harness; FY2026 and FY2025 dollar-based net retention; the $100,000 and $1.0 million ARR customer counts; the patent estate; the securities litigation outcomebacksC01C02C29C37C40C42C46
- S02primarycompany filing2026-09-02the revenue split by product and geography, the restructuring note, the stock-compensation table, remaining performance obligations including the Flex customer advances, and the dollar-based net retention comparisonbacksC01C03C05C13C15C20C21C22C29C30C32C37C39C40C41C42C46
- S03primarycompany filing2026-09-01the income statement, cash flow statement and balance sheet; the GAAP to non-GAAP reconciliation; the Q3 and FY2027 guidance; the buyback figure; the Gartner and Forrester citationsbacksC13C15C17C19C20C22C30C31C32C33
- S04primarytranscript2026-09-01GitLab Q2 FY2027 earnings call transcript, 1 September 2026fmp:earning-call-transcript GTLB 2027Q2management's prepared figures with no filed home - net ARR growth, gross bookings, paid consumption run rate, Flex commitments, sales capacity - and nine analysts' questions, of which one attacked the gross margin directly
Paid consumption run rate ended the quarter >$40 million. Up from $15 million exiting in Q1 thanks to the introduction of Flex. ... roughly 80% of customer query volume comes from customers connecting Orbit to external agents such as Claude code
backsC05C08C09C22C26C28 - S05primarycompany filing2026-06-02the scope and cost of the May 2026 restructuring, and the FY2027 guidance as it stood in JunebacksC33C34
- S06primarycompany filing2026-03-03the original FY2027 guidance issued in March 2026 and the $400 million buyback authorisationbacksC33
- S07primarycompany filing2026-05-11the strategy behind the restructuring, the statement that the majority of savings would be reinvested, and management's own framing of the market moving from tens to hundreds of dollars per user per monthbacksC34
- S08primarycompany filing2026-07-08the FY2025 and FY2026 GAAP to non-GAAP reconciliations, the restated Q1 paid consumption run rate, and the statement that headwinds persist for 20% of ARRbacksC17C26
- S09secondaryvendor data2026-09-08Research pack for GTLB, 8 September 2026content/GTLB/research-pack.jsonprice, returns, the analyst split, short interest from FINRA, insider filings, the earnings surprise history and the filed competitor cohortbacksC19C36C37
- S10primarycompany filing2026-07-29Microsoft's own description of GitHub and the fact that no GitHub revenue line is disclosed anywhere in the filingread, but no claim rests on it
- S11primarycompany filing2026-02-13a competitor naming GitLab and GitHub in the same sentence, and its own account of what home-grown alternatives fail to dobacksC02C09
- S12primarycompany filing2026-08-14the Software Collection product list, and the absence of any mention of GitLab in the whole documentread, but no claim rests on it
- S13primaryindustry data2026-09-08GitHub Docs - about GitHub Copilot coding agent GitHub (Microsoft)the attacker's own statement of what its agent does inside a repository, and that it runs on GitHub Actions
Copilot can research a repository, create an implementation plan, and make code changes on a branch.
backsC06 - S14primaryindustry data2026-09-08GitHub Docs - understand migrations from GitLab to GitHub GitHub (Microsoft)the switching cost stated by the party with every incentive to minimise it: CI/CD pipelines, vulnerability reports, repository policies and group settings are listed as not migrated
Data that GitLab does not include in the export at all, such as webhooks, CI/CD variables, job traces and artifacts, child-pipeline history, and pipeline triggers.
backsC04 - S15primaryindustry data2025-11-05GitHub Octoverse 2025 GitHub (Microsoft)the rate of agent-written change measured on the rival platform, the 180 million developer count and the 36 million who joined in a year
180 million-plus developers now work and build on GitHub ... A first glimpse of coding agent shows 1+ million pull requests that were created between May 2025 and September 2025 ... more than one new developer on average joined GitHub - over 36 million in the past year
backsC07C14 - S16primaryindustry data2026-09-08GitHub pricing page GitHub (Microsoft)a published enterprise seat price from outside GitLab, used as the price factor in the prize
GitHub Enterprise: Starting at $21 USD per user/month
backsC11 - S17primarygovernment data2026-09-08US Bureau of Labor Statistics, Occupational Outlook Handbook - Software Developers, Quality Assurance Analysts, and Testers US Bureau of Labor Statisticsthe buyer population counted by a statistical agency, and the 10% growth projected for 2025-35
Overall employment of software developers, quality assurance analysts, and testers is projected to grow 10 percent from 2025 to 2035, much faster than the average for all occupations. ... Number of Jobs, 2025: 1,905,400
backsC10 - S18primarycompany filing2026-07-01that the chief executive's open-market purchase was executed under a Rule 10b5-1 plan entered into on 25 September 2025, not a discretionary decision at that pricebacksC38
- S19primarycompany filing2026-05-01beneficial ownership and voting power as of 1 April 2026, the dual-class structure, executive biographies, the FY2026 bonus plan's metrics and payout, the PSU design and its forfeiturebacksC43C44C45
- S20primarycompany filing2025-03-21the FY2023 income statement and diluted share count, read through the frozen numbers cardbacksC42C46
- S21primarycompany filing2026-09-01the chief accounting officer's departure on 28 August 2026 and the chief financial officer's assumption of the principal accounting officer rolebacksC45