RuncheyResearch

AI Absorption Ledger / GTLB

GitLab

GTLB · Q3 2026 · reported 2026-09-01 · revenue $286mn

Assessment

Q2 fiscal 2027 is the quarter GitLab widened its consumption metric: platform-wide paid run rate above now counts Flex commitments that customers may fund with former seat dollars, and the agent platform is given only as growth of roughly . Revenue was , up . The ledger reads the agent platform's quarter at and the seat add-ons at ; the calls give inconsistent levels for the end of Q1, and the range spans both.

On cost, management for the first time bounds the AI part of the margin decline: it says the SaaS mix moved the margin more than AI adoption and that many customers pay for inference outside the agreement. The ledger puts model fees and AI compute at , with a ceiling below the SaaS mix's part of the line's excess over its prior-year share. AI engineering is read at net of restructuring charges and the internal AI tools bill at . AI-attributed role savings are left unsized, because AI automation is one of the restructuring's several changes and no source separates its part.

Demand attributed to AI firmed in words: large deals grew more than with AI named among the tailwinds, and seats are extending to non-developers. The channel stays relabelled at . The toll is not mentioned: the segments that hold the price-sensitive cohort stabilized and the call does not name AI as a pressure, so it is left unsized.

Sized channels against the income statement, Q3 2026

7 of 9 channels sized

Each blue mark is one channel's dollars for the quarter; a bar is the range of an estimate. Grey marks are the company's reported lines. The distance between them is the point: how large the AI channel is next to the line it sits in.

New money and old money, Q3 2026

6 new2 expanded1 relabelled

Each channel is tagged once for whether its money existed before language models, from the company's annual report and call at the start of the period. A bar splits one flow's sized dollars by that tag. The incremental total is the part that would not be there without the models: a new channel counts in full, an expanded one only for what AI added, a relabelled one at zero.

Paid for AI$14mn to $55mn sized

new $1.2mn to $7.7mnexpanded $13mn to $47mn

Incremental total $1.2mn to $55mnpoint $3.9mn$30mn in 1 channel has no traced baseline
Revenue arriving through AI$5.0mn to $50mn sized

new $5.0mn to $36mnrelabelled $0 to $15mn

Incremental total $5.0mn to $36mnpoint $14mn

The sized total counts every channel the company credits to AI, including relabelled money that existed before language models and the ledger's own estimates for it. The incremental total counts relabelled channels at zero. A flow is split by layer where its dollars sit at more than one: end use, compute sold to builders, and hardware. The same dollar can be a buyer's spend, a cloud's revenue and a chipmaker's revenue, so the layers are never added together.

Paid for AI3 channels · $14mn to $55mn sized · $1.2mn to $55mn incremental

vendor bill

Model fees and AI compute in cost of revenue

0.39% to 2.2% of the quarter’s revenue

Incremental total: counts in full.

Matched line moved : the whole line, not this channel.

our inferencedisclosure: bounded· motive: product-defensive· before LLMs: new

Asked directly, management says the margin changes come more from the SaaS mix than from AI adoption and that many customers pay for inference outside the GitLab agreement. That bounds the AI part of the line's movement below the SaaS mix's part. The size, , is the ledger's AI revenue estimate times an assumed cost ratio, with a ceiling from that statement applied to the line net of of restructuring charges; the phrase is not in the words table, so the bound enters as an assumption. The counterparty is mixed: third-party model vendors, and the cloud providers that host the AI features.

Evidence: 8 quotes, 5 figures, 2 confounds, 3 from before coverage

What the company pays third-party model vendors and cloud providers to serve its AI features: the seat add-ons, the agent platform including the credits included with every seat and the free-tier access, all inside cost of subscription revenue. Many self-managed customers bring their own models and pay for inference outside the GitLab agreement, so this cost covers only part of the usage.

Why this motive

Carried: inference in cost of revenue for promotional and included credits.

Before LLMs: new

A model and token bill cannot exist without LLMs. At the anchor the AI features already ran on third-party models and the filing warned of high computing costs (claims gtlb-anchor-c6, gtlb-anchor-c7); cost of subscription revenue was , of subscription revenue, with no AI part given.

“Additionally, we have made investments in our cloud provider span to support our AI and R&D efforts.”
CFO, prepared remarks, earnings call, 2023-06-05
“We rely on third-party vendors for the provision of the AI models which power many of our AI features.”
Filing, risk factors, 10-K periodic report, 2024-03-26
“Further, developing, testing, and offering AI-powered features may lead to greater than expected expenditures for our company because deploying AI systems involves high computing costs, which could adversely affect our gross margin, profitability, financial position, and cash flow.”
Filing, risk factors, 10-K periodic report, 2024-03-26

Figures

  • Non-GAAP gross margin · 2026-CQ3
  • Increase in third-party hosting costs in cost of revenue, year over year · 2026-CQ3
  • Restructuring charges in cost of revenue · 2026-CQ3
  • Cost of subscription revenue less all restructuring charges in cost of revenue · 2026-CQ3
  • Cost of subscription revenue, net of restructuring, above its prior-year share of subscription revenue · 2026-CQ3

Reported line it is matched to

Cost of subscription revenue was against , above its prior-year share of subscription revenue. The filing attributes the rise and the gross margin decline to third-party hosting for SaaS and cloud usage, an increase of , and does not name AI.

2026-CQ3: 2025-CQ3: 2026-CQ3: 2025-CQ3:

What else could explain it

  • mix shift: SaaS and Dedicated growth, which management names as the larger cause.
  • one time item: Restructuring costs are in cost of revenue this quarter.

Quotes

“What that means is, for many of our customers, the token or inference cost is actually not embedded in the GitLab agreement.”
c11 · CEO, qa, earnings call, 2026-09-01
“They pay us for the access to the platform, and they pay for the work done in the platform, the context, the harness, the governance and auditability that we provide, not the inference. Those are all very high-margin products.”
c37 · CEO, qa, earnings call, 2026-09-01
“As Jessica alluded to, a lot of the margin changes that we've seen in the business have been driven more by the mix shift to SaaS than the early AI adoption.”
c12 · CEO, qa, earnings call, 2026-09-01
“We have intentionally invested in consumption products and pushing customers to focus on transitioning pilots to production.”
c13 · CFO, qa, earnings call, 2026-09-01
“Non-GAAP gross margin was 86.5%. SaaS was 34% of total revenue and grew 36% year-over-year, powered by continued strength in GitLab Dedicated and Duo.”
c14 · CFO, prepared remarks, earnings call, 2026-09-01
“Cost of revenue increased by $17.1 million, to $45.6 million for the three months ended July 31, 2026 from $28.5 million for the three months ended July 31, 2025, primarily due to an increase of $11.0 million in third party hosting costs for SaaS and cloud usage.”
c15 · Filing, mdna, 10-Q periodic report, 2026-09-02
“The decrease in gross margin was primarily attributable to an increase in third party hosting costs for SaaS and cloud usage.”
c16 · Filing, mdna, 10-Q periodic report, 2026-09-02
“Further, developing, testing, and offering AI-powered features may lead to greater than expected expenditures for our company because deploying AI systems involves high computing costs, which could adversely affect our gross margin, profitability, financial position, and cash flow.”
c17 · Filing, risk factors, 10-Q periodic report, 2026-09-02

By quarter

  • Q1 2026direction only · our inference · product-defensive · $849k to $7.1mn
  • Q2 2026direction only · our inference · product-defensive · $918k to $9.2mn
  • Q3 2026bounded · our inference · product-defensive · $1.1mn to $6.3mn

engineering

Engineering and adoption spending on the agent platform and AI capabilities

4.5% to 16.4% of the quarter’s revenue

Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.

our inferencedisclosure: described· motive: exploratory· before LLMs: expanded

Research and development was , including of restructuring charges, against . The Git rebuild is under way and the Orbit context service is in public beta, enabled by more than organizations. The size, , is an assumed share of the line net of restructuring.

Evidence: 5 quotes, 4 figures, 2 confounds, 2 from before coverage

Research and development directed to AI: the agent platform, the Orbit context service, the rebuild of Git for agent-rate load, and the forward-deployed engineers and technical services that get customers' agent work into production. Sized from research and development; adoption staff in sales and services lines are named and not added.

Why this motive

Carried: an investment year; the new products are in beta (claims c25, c13).

Before LLMs: expanded

Engineering budget redirected to AI is in the anchor: engineers were moved from other teams to AI work without significant added expense, and cloud spending supported AI and research (claims gtlb-anchor-c3, gtlb-anchor-c4). Research and development was in fiscal 2024; no AI part is given, so there is no traced baseline for the AI share. The size is the whole of an activity that existed before.

“While we have had some teams working on AI features, we recently shifted additional engineers from other teams to support the work on AI. As a result, this has not led to significant incremental expenses on engineering talent.”
CFO, prepared remarks, earnings call, 2023-06-05
“Additionally, we have made investments in our cloud provider span to support our AI and R&D efforts.”
CFO, prepared remarks, earnings call, 2023-06-05

Figures

  • Research and development · 2026-CQ3
  • Research and development, prior-year quarter · 2025-CQ3
  • Restructuring charges in research and development · 2026-CQ3
  • Organizations that enabled Orbit indexing, floor · as-of 2026-09-01

What else could explain it

  • one time item: Restructuring charges and an in-person company event are in the line.
  • line composition: Research and development funds every product; the AI share is assumed.

Quotes

“We're now in the middle of re-architecting that Git infrastructure to achieve roughly 100x scale what humans have ever required.”
c24 · CEO, qa, earnings call, 2026-09-01
“Since opening the beta in June, more than 2,200 organizations have enabled Orbit indexing, an increase of 70% in four weeks.”
c25 · CEO, prepared remarks, earnings call, 2026-09-01
“Expanded context for AI agents with GitLab Orbit, available in public beta, by connecting code, work items, pipelines, deployments, and production signals into a unified context graph”
c26 · Filing, press release, 8-K earnings release, 2026-09-01
“Research and development expenses increased by $23.5 million, to $95.0 million for the three months ended July 31, 2026 from $71.5 million for the three months ended July 31, 2025, primarily driven by an increase of $8.0 million in restructuring costs”
c27 · Filing, mdna, 10-Q periodic report, 2026-09-02
“We have intentionally invested in consumption products and pushing customers to focus on transitioning pilots to production.”
c13 · CFO, qa, earnings call, 2026-09-01

By quarter

  • Q1 2026described · our inference · exploratory · $10mn to $34mn
  • Q2 2026described · our inference · exploratory · $11mn to $36mn
  • Q3 2026described · our inference · exploratory · $13mn to $47mn

vendor bill

Internal AI tools bill

0.03% to 0.47% of the quarter’s revenue

Incremental total: counts in full.

Matched line moved : the whole line, not this channel.

our inferencedisclosure: described· motive: narrative-defensive· before LLMs: new

The filing again names spending on AI tools. Software expenses rose in research and development and in general and administrative, without attribution. The size, , is team members after the restructuring times an assumed adoption share and price; the software increase is a cross-check on the change, not a ceiling on the level.

Evidence: 3 quotes, 2 figures, 1 confound

What the company pays for AI tools its own staff use: coding agents and assistants for engineers and AI tooling across functions, subject by the filing's account to third-party price increases, usage-based variability and contractual minimums. Management plans daily AI use by every employee and puts part of the restructuring savings into internal AI tooling.

Why this motive

Carried: a usage mandate.

Before LLMs: new

An AI tools bill cannot exist without LLMs. Neither anchor source mentions AI tools for the company's own staff; no dollar baseline exists.

Figures

  • Team members after the restructuring, approximate · as-of 2026-07-31
  • Increase in software expenses in research and development and general and administrative, year over year · 2026-CQ3

Reported line it is matched to

Software expenses in research and development and in general and administrative rose together; the filing does not say what software, and a year-over-year rise does not bound the level.

2026-CQ3: 2026-CQ3:

What else could explain it

  • line composition: Software expenses include every subscription the company pays for.

Quotes

“In addition, we have made, or intend to make, investments in AI tools whose costs are subject to third-party pricing increases, usage-based variability, and contractual minimums and if these AI tooling costs increase, our operating expenses and cash flow could be adversely affected.”
c30 · Filing, risk factors, 10-Q periodic report, 2026-09-02
“The remaining change was primarily attributable to an increase of $3.6 million in hosting expenses and an increase of $1.6 million in software expenses.”
c28 · Filing, mdna, 10-Q periodic report, 2026-09-02
“an increase of $5.2 million in restructuring costs, and an increase of $1.3 million in software expenses.”
c29 · Filing, mdna, 10-Q periodic report, 2026-09-02

By quarter

  • Q2 2026described · our inference · narrative-defensive · $46k to $1.4mn
  • Q3 2026described · our inference · narrative-defensive · $80k to $1.3mn

Cost displaced by AI1 channel · 1 not sized

back office · cheap to verify

Roles and process cost removed by internal AI agents

Not sized

AI automation of internal reviews, approvals and handoffs is one of several operational changes in the restructuring, beside fewer countries, fewer management layers and smaller research teams (Q2 claims c39 and c40), and no source gives AI's share of the affected roles (the methodology rule for AI named beside another cause).

described, no sizedisclosure: described· motive: narrative-defensive· before LLMs: expanded

The restructuring took place at the start of the quarter, affecting about team members, with charges of about . The 10-Q expects separations to be substantially complete by fiscal year end and shows paid against of severance charges. The quarter's sources do not return to the AI part of it. AI automation is one of the restructuring's several changes and nothing separates its part, so the channel stays described and unsized, as in Q2.

Evidence: 3 quotes, 4 figures, 2 confounds, 1 from before coverage

Personnel cost the company no longer pays because internal reviews, approvals and handoffs are automated with AI agents and roles are right-sized to follow, one of the operational changes in the restructuring announced in May 2026. Management says the restructuring is not an AI optimization or cost-cutting exercise and reinvests most of the savings; the other changes (fewer countries, fewer management layers, smaller research teams) are confounds and nothing of them is credited here.

Why this motive

Carried: AI-attributed role reductions that management says are not an AI cost-cutting exercise.

Before LLMs: expanded

The work and its cost are in the anchor: the company had team members at the end of fiscal 2024 (claim gtlb-anchor-c10), and no AI was used in internal processes then by the anchor's account. The size is the change AI made, not the whole line.

“As of January 31, 2024, we had approximately 2,130 team members in 65 countries.”
Filing, business, 10-K periodic report, 2024-03-26

Figures

  • Restructuring charges in the quarter, approximate · 2026-CQ3
  • Team members affected by the restructuring · as-of 2026-06-02
  • Restructuring charges under ASC 420, six months · 2026-02-01..2026-07-31
  • Restructuring cash paid under ASC 420, six months · 2026-02-01..2026-07-31

What else could explain it

  • transformation program: Fewer countries, fewer management layers and smaller research teams; nothing of their saving is credited to AI.
  • one time item: Restructuring charges raise every operating line this quarter.

Quotes

“We incurred approximately $23.3 million in restructuring charges, in line with what we outlined last quarter.”
c31 · CFO, prepared remarks, earnings call, 2026-09-01
“I especially also want to recognize the entire GitLab team, because at the beginning of this quarter, we made the difficult decision to restructure the company.”
c32 · CEO, prepared remarks, earnings call, 2026-09-01
“The 2027 Plan impacted approximately 14% of the Company’s global workforce and is expected to be substantially complete with respect to employee separations by the end of fiscal year 2027.”
c33 · Filing, notes, 10-Q periodic report, 2026-09-02

By quarter

  • Q2 2026described · described, no size · narrative-defensive
  • Q3 2026described · described, no size · narrative-defensive

Revenue arriving through AI4 channels · $5.0mn to $50mn sized · $5.0mn to $36mn incremental

product revenue · cheap to verify

Duo Agent Platform credits (usage-based agent work)

1.3% to 2.2% of the quarter’s revenue

Incremental total: counts in full.

our inferencedisclosure: quantified· motive: exploratory· before LLMs: new

The run rate metric widened this quarter. The stated level is now platform-wide paid run rate above , which includes Flex commitments that may fund seats, against a restated for the end of Q1; that is below the agent platform figure of given then for the same date. The agent platform's own run rate is given as growth of roughly , now including Flex reservations, and the channel is sized from that rate. The size, , spans both opening levels.

Evidence: 14 quotes, 8 figures, 2 confounds, 2 from before coverage

Revenue from GitLab Duo Agent Platform, generally available in January 2026 and priced in GitLab Credits: committed monthly credit pools recognized ratably, on-demand credits billed on usage, and from mid-2026 agent reservations inside Flex commitments. Management reports a paid consumption run rate at period ends; no income-statement line carries it. The work automated is code review, failed-pipeline fixes, dependency updates and vulnerability remediation, each submitted as a merge request a human reviews.

Why this motive

Carried: management still assumes limited contribution this year (claim c5); customers spending multiples of seat price on credits is an offensive tell, and the less durable motive is kept.

Before LLMs: new

Priced units of agent work did not exist before LLMs. At the anchor GitLab Duo is a suite of AI features and consumption pricing covers only compute and storage (claims gtlb-anchor-c1, gtlb-anchor-c9). No agent product or credit was offered; no dollar baseline exists.

“Our suite of AI capabilities, known as GitLab Duo, improves every step of software development and delivery – from planning and code generation to vulnerability detection and value stream measurement.”
Filing, business, 10-K periodic report, 2024-03-26
“We already have consumptive elements in our model. For example, for compute and for storage, you pay on a consumption basis.”
CEO, qa, earnings call, 2023-06-05

Figures

  • Duo Agent Platform consumption run rate, quarter-over-quarter growth, approximate · 2026-CQ3
  • Platform-wide paid consumption run rate at quarter end, floor · as-of 2026-07-31
  • Platform-wide paid consumption run rate at the end of the prior quarter, as restated on this call · as-of 2026-04-30
  • Paid consumption run rate target for the end of fiscal 2027 · as-of 2027-01-31
  • Flex commitments, cumulative since the June 2026 launch, floor · as-of 2026-07-31
  • Customers with Flex commitments, floor · as-of 2026-07-31
  • Flex commitments not yet provisioned to any product · as-of 2026-07-31
  • Flex commitments billed and held as customer advances · as-of 2026-07-31

What else could explain it

  • relabel: The run rate metric widened: it now counts Flex commitments, which customers may fund with former seat dollars, and the agent platform figure counts Flex reservations. The 10-Q puts Flex commitments not yet provisioned to any product at , of which had been billed, so much of the platform-wide level is unallocated commitment. Read on this quarter's definition; not backfilled.
  • relabel: Seat add-on contracts converted during the year move into this channel.

Quotes

“Paid CRR ended the quarter above $40 million, up from $15 million existing in Q1, thanks to the introduction of Flex.”
c1 · CEO, prepared remarks, earnings call, 2026-09-01
“As a reminder, paid CRR is a point-in-time annualized measure that includes GitLab Credit commitments, Flex commitments, and paid on-demand consumption. It excludes trials and promotional credits.”
c2 · CEO, prepared remarks, earnings call, 2026-09-01
“This quarter, Duo Agent Platform CRR grew roughly 50% quarter-over-quarter, inclusive of credit commitments, paid on-demand credits, and Flex reservations.”
c3 · CEO, prepared remarks, earnings call, 2026-09-01
“Our objective is to exceed $100 million of paid CRR by the end of this fiscal year.”
c4 · CEO, prepared remarks, earnings call, 2026-09-01
“Third, on Duo Agent Platform, we assume limited contribution in FY 2027 relative to our large existing revenue base.”
c5 · CFO, prepared remarks, earnings call, 2026-09-01
“We have got several now that are spending multiples in excess of their Premium or Ultimate seat price now in terms of credits.”
c6 · CEO, qa, earnings call, 2026-09-01
“We also broadened adoption geographically after a more U.S.-centric first quarter, and one top 20 U.S. commercial bank expanded its AI credit pool nearly tenfold this quarter.”
c7 · CEO, prepared remarks, earnings call, 2026-09-01
“After only six weeks in market, more than 130 customers committed more than $20 million to Flex.”
c8 · CEO, prepared remarks, earnings call, 2026-09-01
“Clearly, customers see the benefit of it, and they may start to fund their Flex agreement with their previous seat-based subscription dollars.”
c9 · CEO, qa, earnings call, 2026-09-01
“This is the first quarter Flex shows up in our results, and the impact was therefore immaterial relative to the size of our existing revenue base.”
c10 · CFO, prepared remarks, earnings call, 2026-09-01
“What that means is, for many of our customers, the token or inference cost is actually not embedded in the GitLab agreement.”
c11 · CEO, qa, earnings call, 2026-09-01
“They pay us for the access to the platform, and they pay for the work done in the platform, the context, the harness, the governance and auditability that we provide, not the inference. Those are all very high-margin products.”
c37 · CEO, qa, earnings call, 2026-09-01
“Quantified the potential business value of GitLab Duo Agent Platform through an independent Forrester Consulting Total Economic Impact™ study, which found organizations can achieve a 400% return on investment”
c39 · Filing, press release, 8-K earnings release, 2026-09-01
“of which $40.4 million relates to non-cancellable customer advances under Flex contracts where specific products and quantities are determined by the customer at a later date, of which $15.0 million had been billed and was reflected in customer advances as of that date”
c34 · Filing, mdna, 10-Q periodic report, 2026-09-02

By quarter

  • Q1 2026described · our inference · exploratory · $25k to $750k
  • Q2 2026quantified · our inference · exploratory · $1.9mn to $3.3mn
  • Q3 2026quantified · our inference · exploratory · $3.7mn to $6.4mn

product revenue · cheap to verify

Duo Pro and Duo Enterprise seat add-ons

0.47% to 10.2% of the quarter’s revenue

Incremental total: counts in full.

our inferencedisclosure: described· motive: offensive· before LLMs: new

Duo is credited, with Dedicated, for SaaS growth of ; the add-ons are otherwise not discussed as contracts move into the agent platform. The size, , is the ledger's inference, lowered for conversions.

Evidence: 1 quote, 1 figure, 2 confounds, 2 from before coverage

Seat-priced AI add-ons sold to Premium and Ultimate customers: code completion, generation and chat (Duo Pro) and, in Duo Enterprise, vulnerability analysis, pipeline root-cause analysis and self-hosted models. Management is converting these contracts into agent platform credits during fiscal 2027; a converted contract leaves this channel for the agent platform channel.

Why this motive

Carried: a separately priced AI add-on.

Before LLMs: new

A separately priced LLM product. At the anchor it was a plan: the first AI add-on, with Code Suggestions in beta, was to be priced at per user per month (claim gtlb-anchor-c2). The product is an LLM product throughout, so the anchor's mention does not make it older money; no revenue was disclosed then or since.

“Our suite of AI capabilities, known as GitLab Duo, improves every step of software development and delivery – from planning and code generation to vulnerability detection and value stream measurement.”
Filing, business, 10-K periodic report, 2024-03-26
“Later this year, we plan to introduce an AI add-on focused on supporting development teams. This new add-on will include Code Suggestions functionality. We anticipate this will be priced at $9 per user per month billed annually.”
CEO, prepared remarks, earnings call, 2023-06-05

Figures

  • SaaS revenue, year-over-year growth · 2026-CQ3

What else could explain it

  • relabel: Converted contracts leave this channel for the agent platform channel.
  • line composition: SaaS growth is credited to Dedicated and Duo together.

Quotes

“Non-GAAP gross margin was 86.5%. SaaS was 34% of total revenue and grew 36% year-over-year, powered by continued strength in GitLab Dedicated and Duo.”
c14 · CFO, prepared remarks, earnings call, 2026-09-01

By quarter

  • Q1 2026described · our inference · offensive · $1.8mn to $28mn
  • Q2 2026described · our inference · offensive · $2.0mn to $30mn
  • Q3 2026described · our inference · offensive · $1.4mn to $29mn

customer cohort

Core seat and platform demand attributed to AI-driven code volume

0% to 5.2% of the quarter’s revenue

Incremental total: counts at zero.

our inferencedisclosure: direction only· motive: exploratory· before LLMs: relabelled

Management says customers that adopt AI tools use more of the platform, with code pushes up and pipelines up , and credits AI in part for growth of more than in large deals, beside public sector and execution. The filing does not name AI. The size, , is an assumed share of subscription growth of after taking out the agent platform and seat add-on revenue counted in their own channels.

Evidence: 8 quotes, 4 figures, 2 confounds, 1 from before coverage

Ordinary Premium, Ultimate and Dedicated subscription revenue that management says arrives because AI raises code volume, pipelines and the number of people who write code (non-engineers given seats to contribute with agentic tools). Distinct from the AI products themselves.

Why this motive

Carried (exploratory): AI is credited with demand through activity rates and, in part, with large-deal growth beside public sector and execution (claims c19, c21); with no revenue measure the exploratory tell holds, and no narrative-defensive tell of its own is quoted.

Before LLMs: relabelled

Seat subscriptions are the business that predates agents: subscription revenue was in fiscal 2024. On the anchor call the CEO already argued that generative AI would bring more coders and more code onto the platform (claim gtlb-anchor-c5). The same argument, with no line attributed to it, is read as a new name for existing demand.

“Yeah, we believe that generative AI will expand the market. First of all, you make the product easier, like coding today is hard, and AI makes it easier. We expect these citizen developers, these junior developers, to start coding. That code needs to be managed somewhere, and that is in GitLab.”
CEO, qa, earnings call, 2023-06-05

Figures

  • Code pushes, year-over-year growth · 2026-CQ3
  • CI/CD pipelines, year-over-year growth · 2026-CQ3
  • Deals of $500,000 or more, year-over-year growth, floor · 2026-CQ3
  • Subscription revenue, year-over-year increase · 2026-CQ3

What else could explain it

  • other: Public sector rebounded and sales capacity and productivity rose, by management's account.
  • mix shift: Ultimate grew on security and compliance demand.

Quotes

“Year-over-year, secure repositories grew 60%, code pushes grew 50%, CI/CD pipelines grew 40%. Among some customers moving aggressively into AI-assisted development, we have seen code bases grow as much as 500%. The pattern is increasingly clear. As enterprises adopt more AI development tools, they use more of GitLab.”
c18 · CEO, prepared remarks, earnings call, 2026-09-01
“We also see AI tailwinds, though, helping drive large deals. Jessica mentioned our $500K and above deals grew more than 150% year-over-year.”
c19 · CEO, qa, earnings call, 2026-09-01
“We are starting to see demand extend beyond the traditional developer seat entirely as AI makes software creation accessible to a much broader set of builders across the enterprise.”
c20 · CFO, prepared remarks, earnings call, 2026-09-01
“I think it is really a result of many quarters of investment and some tailwinds kicking in across multiple dimensions. As Jessica mentioned, PubSec rebounded meaningfully.”
c21 · CEO, qa, earnings call, 2026-09-01
“In terms of AI natives, what I would say there is really this is an opportunity. They are creating tailwinds for us. They have dramatically simplified the ability for anyone to create code, and all of that code needs GitLab.”
c22 · CEO, qa, earnings call, 2026-09-01
“One of the things that's also inside of that more new customers number is really an increased demand for seats. AI has significantly enabled anyone to become a builder.”
c35 · CEO, qa, earnings call, 2026-09-01
“We are now seeing, as I have mentioned, our customers coming to us and saying they need more GitLab seats because of non-engineering people who need access to the platform.”
c36 · CEO, qa, earnings call, 2026-09-01
“As AI drives more software creation and more work through the development lifecycle, the context, security, governance and control GitLab provides become increasingly valuable.”
c38 · Filing, press release, 8-K earnings release, 2026-09-01

By quarter

  • Q1 2026described · described, no size · exploratory
  • Q2 2026direction only · described, no size · exploratory
  • Q3 2026direction only · our inference · exploratory · $0 to $15mn

customer cohort

Subscriptions sold to AI labs and AI start-ups

0.03% to 2.6% of the quarter’s revenue; overlaps another channel, not added into totals

our inferencedisclosure: described· motive: exploratory· before LLMs: new

AI start-ups are named among the small first orders, which more than doubled. No revenue or count is given. The size, , rests on assumed counts and contract sizes.

Evidence: 1 quote, 1 confound

GitLab subscriptions bought by AI labs and AI start-ups, which management names as customers. One lab is also a design partner on the rebuild of Git. Kept apart from enterprise revenue because these payers are funded largely by investors; their AI-driven growth may also sit in the core demand channel, so the readings list that channel as an overlap and stay out of totals.

Why this motive

Carried from Q2 (exploratory): AI start-ups are named among small first orders (claim c23) with no revenue, count or reason for their spend, and the sources do not contradict each other.

Before LLMs: new

The payers exist because of LLMs. Neither anchor source names an AI lab or AI start-up as a customer, and no revenue from them has been disclosed.

What else could explain it

  • other: Small first orders carry little revenue in their first year, by management's account.

Quotes

“It is important to win customers of all sizes, including AI startups and others with small orders to begin with, because more than half of our current 1 million-plus in run rate revenue comes from customers whose first order was less than $5,000.”
c23 · CEO, prepared remarks, earnings call, 2026-09-01

By quarter

  • Q2 2026described · our inference · exploratory · $100k to $7.3mn
  • Q3 2026described · our inference · exploratory · $100k to $7.5mn

Cost imposed, or revenue lost, by others’ AI1 channel · 1 not sized

customer cohort

Premium growth lost to customers' AI code experimentation

Not sized

Management is silent on AI as a cause of Premium pressure this quarter; with no claim to apply a share to, the channel is left unsized.

inscrutabledisclosure: not mentioned· motive: imposed· before LLMs: new

The quarter's sources do not attribute any pressure to customers' AI. The call reports that the small and mid-market segments, which hold the price-sensitive cohort, stabilized ahead of targets, without naming AI; the 10-Q repeats a general risk factor on total revenue, which is not a reading of this channel.

Evidence: 0 quotes, 1 from before coverage

Seat and tier revenue the company does not receive because price-sensitive customers put budget into AI coding tools instead of Premium seats. Management names rising AI code experimentation as one cause, beside a Premium price increase, flat SaaS budgets and its own upmarket shift, of slower Premium growth in its price-sensitive cohort.

Why this motive

Carried from the prior quarter: a toll, not the company's choice.

Before LLMs: new

Revenue lost to customers' AI tools could not exist before them. Seat contraction is in the anchor, attributed to lower seat counts under macro pressure and not to AI (claim gtlb-anchor-c8); no dollar size was given then.

“Contraction improved over Q4, but is higher than prior quarters. Like Q4, contraction is driven almost entirely by lower seat counts with minimal downtiering.”
CFO, prepared remarks, earnings call, 2023-06-05

By quarter

  • Q1 2026described · described, no size · imposed
  • Q2 2026described · described, no size · imposed
  • Q3 2026not mentioned · inscrutable · imposed

Reported lines, year-over-year growth

Revenue +21.3%

Q3 2026. No tracked cost line grew slower than revenue. A displaced cost shows up as a line that stays under the dashed revenue line. These are the audited lines, as first reported; nothing here is attributed to AI by the filing. Not drawn: cost of revenue, which one-time items move by more than 60% in a quarter; the values are in the table below.

Research and developmentSales and marketingGeneral and administrativeRevenue
-20%0%20%40%60%Q1 2025Q2 2025Q3 2025Q4 2025Q1 2026Q2 2026Q3 2026General and administrativeResearch and developmentSales and marketingRevenue
Reported values and filings
LineQ2 2025Q3 2025Q4 2025Q1 2026Q2 2026Q3 2026
Revenue$215mn$236mn$244mn$260mn$264mn$286mn
Cost of revenue$25mn$29mn$32mn$35mn$37mn$46mn
Research and development$65mn$71mn$69mn$69mn$71mn$95mn
Sales and marketing$108mn$110mn$105mn$113mn$119mn$134mn
General and administrative$51mn$45mn$51mn$49mn$52mn$68mn