GitHub Statistics 2026
By Axis Intelligence Research
Co-author: Sarah Mitchell | Last updated: September 22, 2026 | License: CC BY 4.0
GitHub reached 225 million users as of July 29, 2026, according to Microsoft chairman and chief executive Satya Nadella on the company’s fiscal fourth-quarter earnings call. That is 45 million more accounts than the 180 million GitHub reported in its Octoverse 2025 research year, which closed on August 31, 2025. GitHub Copilot now counts 50 million users, and one in three pull requests on the platform involves an agent.
Quick Answer: How many people use GitHub in 2026?
GitHub has 225 million users (Microsoft, July 29, 2026) and 50 million GitHub Copilot users. According to Axis Intelligence Research, that puts the Copilot attach rate at 22.2% of all GitHub accounts. Across the economies GitHub reports publicly, developer accounts grew 30.7% year over year in Q1 2026, and India now stands at 82.5% of the United States developer base, up from 69.7% in Q2 2024.
Key Findings
- According to Axis Intelligence Research, GitHub Copilot’s attach rate reached 22.2% of all GitHub accounts in July 2026, derived from Microsoft’s two disclosed counts of 50 million Copilot users and 225 million GitHub users.
- According to Axis Intelligence Research, India’s GitHub developer base equalled 82.5% of the United States total in Q1 2026, up from 74.5% one year earlier and 69.7% in Q2 2024.
- According to Axis Intelligence Research, South Korea ranks first on the Axis Developer Capacity Index with a score of 72.3, ahead of the United States at 61.5, on the strength of the highest organizational density of any large economy at 81.9 organizations per 1,000 developers.
- According to Axis Intelligence Research, developers in the United States held 18.9 GitHub accounts for every one employed US software developer in 2025, combining GitHub Innovation Graph account counts with Bureau of Labor Statistics occupational employment.
- According to Axis Intelligence Research, the repository topic
mcpgrew from zero reported pushers in Q1 2024 to 10,500 in Q1 2026, the fastest standing-start adoption of any topic tracked in GitHub’s public dataset.
How Many Developers Are on GitHub in 2026?
Two numbers matter, and they come from different places.
The platform-wide count is Microsoft’s. On July 29, 2026, Nadella told analysts that GitHub had reached 225 million users and that more than 90% of the Fortune 500 build on it. That figure sits in the Microsoft FY26 Q4 earnings call transcript and is the freshest platform total on record.
The structural count is GitHub’s own, published annually in Octoverse. The Octoverse 2025 report covered September 1, 2024 through August 31, 2025 and put the platform at more than 180 million developers, with 36 million joining inside that single year. GitHub called it the fastest absolute growth in its history and attributed the step change to the December 2024 launch of Copilot Free.
Between those two dates, roughly eleven months apart, GitHub added about 45 million accounts. According to Axis Intelligence Research, that is a 25.0% increase against the Octoverse base, and both counts use the same underlying definition: a developer is anyone with a GitHub account.
GitHub platform totals, 2025 to 2026
| Metric | Value | As of | Source |
|---|---|---|---|
| GitHub users | 225,000,000 | Jul 29, 2026 | Microsoft FY26 Q4 call |
| GitHub developers | 180,000,000+ | Aug 31, 2025 | GitHub Octoverse 2025 |
| New developers, Octoverse year | 36,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Total repositories | 630,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Public repositories | 395,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| New repositories in the year | 121,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Public contributions | 1,120,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Private contributions | 4,970,000,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Pull requests merged, monthly average | 43,200,000 | Aug 31, 2025 | GitHub Octoverse 2025 |
| Fortune 500 using GitHub | Over 90% | Jul 29, 2026 | Microsoft FY26 Q4 call |
Why a GitHub account is not a software job
Account totals get quoted as if they were a census of working programmers. They are not, and the gap is instructive rather than embarrassing.
The US Bureau of Labor Statistics counted 1,717,800 employed software developers in the United States in 2025, projected to reach 1,892,600 by 2035, a 10% decade of growth against 3% for all occupations. Median pay was $135,980 in May 2025. GitHub’s public data put US developer accounts at 32,410,529 in Q1 2026.
According to Axis Intelligence Research, that is 18.9 GitHub accounts for every employed US software developer. The ratio is a scale check, not a duplication rate: it folds together students, hobbyists, data scientists, researchers, designers, second accounts, dormant accounts and the large population of people who write code without holding the BLS occupation code. Read it as the distance between platform reach and payroll, which is the number that actually matters when a vendor quotes 225 million users as a market.
Sarah Mitchell: Every platform in this category reports the metric that flatters it, and account totals are the softest of them. The BLS number is a headcount of people paid to do the job under a specific classification. GitHub’s is a count of registrations. Neither is wrong. The mistake is treating one as a proxy for the other, which is exactly what happens when a slide says “180 million developers” and a budget owner hears “addressable engineers.”
Which Countries Have the Most GitHub Developers?
For country-level numbers, the useful source is not Octoverse but the GitHub Innovation Graph, a quarterly public dataset released under CC0 that reports developers, git pushes, repositories, organizations, languages, licenses and topics for every economy clearing a 100-developer privacy threshold. The most recent quarter available is Q1 2026, which makes it roughly two quarters fresher than the annual Octoverse cut. Locations are assigned by IP address using each account’s mode daily location.
Top 20 economies by GitHub developer accounts, Q1 2026
| Rank | Economy | Developers | YoY growth | Git pushes (quarter) | Pushes per developer | Organizations |
|---|---|---|---|---|---|---|
| 1 | United States | 32,410,529 | +25.4% | 57,234,005 | 1.77 | 1,693,029 |
| 2 | India | 26,727,004 | +38.8% | 44,658,960 | 1.67 | 764,137 |
| 3 | China | 11,351,939 | +9.3% | 2,917,473 | 0.26 | 351,045 |
| 4 | Brazil | 8,317,960 | +36.4% | 13,510,410 | 1.62 | 356,493 |
| 5 | United Kingdom | 5,559,767 | +24.5% | 10,466,194 | 1.88 | 283,126 |
| 6 | Japan | 5,433,912 | +36.5% | 10,424,896 | 1.92 | 267,687 |
| 7 | Indonesia | 5,270,032 | +32.4% | 6,880,934 | 1.31 | 228,793 |
| 8 | Germany | 5,255,625 | +31.2% | 11,508,698 | 2.19 | 287,945 |
| 9 | Russia | 4,627,668 | +16.8% | 5,343,534 | 1.15 | 132,349 |
| 10 | France | 4,034,001 | +27.9% | 9,731,787 | 2.41 | 213,516 |
| 11 | Canada | 3,991,305 | +24.3% | 8,225,761 | 2.06 | 243,463 |
| 12 | Hong Kong SAR | 3,485,551 | +36.1% | 4,449,662 | 1.28 | 146,253 |
| 13 | South Korea | 3,145,730 | +29.2% | 9,438,093 | 3.00 | 257,666 |
| 14 | Vietnam | 2,972,715 | +35.2% | 5,146,573 | 1.73 | 158,105 |
| 15 | Mexico | 2,865,151 | +35.1% | 3,192,114 | 1.11 | 92,440 |
| 16 | Philippines | 2,815,527 | +41.3% | 3,791,312 | 1.35 | 67,164 |
| 17 | Spain | 2,802,085 | +32.4% | 6,265,787 | 2.24 | 136,770 |
| 18 | Turkiye | 2,544,065 | +30.2% | 3,421,863 | 1.35 | 113,975 |
| 19 | Pakistan | 2,497,496 | +45.3% | 3,523,185 | 1.41 | 74,034 |
| 20 | Australia | 2,368,357 | +26.6% | 3,736,152 | 1.58 | 126,191 |
Source: GitHub Innovation Graph, Q1 2026. Growth rates and pushes-per-developer calculated by Axis Intelligence Research.
Three things jump out of that table that no headline count shows.
China ranks third by accounts and sixty-eighth by push intensity, at 0.26 pushes per developer against a top-20 median of 1.65. Some of that is genuine platform substitution toward domestic hosting, and some is measurement: GitHub assigns activity by IP address, and the Innovation Graph documentation states plainly that VPN use distorts economy-level metrics. Both readings point the same way, which is that Chinese account totals should not be read as Chinese activity totals.
Russia and Ukraine are the only large economies growing under 21%, at 16.8% and 20.9%.
And the fastest large-economy growth is not in the usual places. Pakistan at 45.3%, the Philippines at 41.3%, Egypt at 43.9% and South Africa at 42.8% all outgrew India, from smaller bases.
When will India overtake the United States on GitHub?
GitHub’s own forecast, published in Octoverse 2025, puts India at 57.5 million developers by 2030 and the United States at 54.7 million, without naming a crossover date. The quarterly data supports a considerably earlier one.
Over the eight quarters from Q2 2024 to Q1 2026, India’s developer base compounded at 7.57% per quarter and the United States at 4.97%. Fitting ordinary least squares to the logarithm of each series and solving for equality puts the crossover in Q4 2027 to Q1 2028, at roughly 46.5 million developers each. According to Axis Intelligence Research, that is about two years ahead of what GitHub’s published 2030 figures imply.
The formula, stated so anyone can rerun it: ln(developers) = a + b x quarter_index, fitted separately for each economy on the eight quarters ending Q1 2026, then solve a_IN + b_IN x t = a_US + b_US x t.
That estimate is sensitive to functional form, and publishing the sensitivity is more useful than hiding it. Fitting the same eight quarters in absolute counts rather than logarithms pushes the crossover to mid-2033. Axis Intelligence Research adopts the compound model as the headline estimate because both series are compounding, not adding fixed increments, and a linear fit on a compounding series systematically understates the faster grower. Readers who prefer the conservative reading have the alternative number and the method to check it.
| Quarter | India as % of US developers |
|---|---|
| Q2 2024 | 69.7% |
| Q1 2025 | 74.5% |
| Q1 2026 | 82.5% |
Source: Axis Intelligence Research calculation on GitHub Innovation Graph developer counts.
The Axis Developer Capacity Index (DCI): Which Economies Punch Above Their Size?
Raw developer counts reward population. They tell you almost nothing about whether an economy’s developer base is shipping code, organizing into teams, or standing still.
The Developer Capacity Index (DCI) is an Axis Intelligence Research metric that scores an economy’s GitHub ecosystem on four components drawn entirely from GitHub’s public quarterly dataset, so that any reader can recompute it from the same CC0 files.
Components and weights
| Component | What it measures | Input | Weight |
|---|---|---|---|
| Scale | Size of the developer base | log10(developers), Q1 2026 | 30% |
| Intensity | Whether accounts ship code | git pushes per developer, Q1 2026 | 25% |
| Organizational depth | Whether developers cluster into teams, companies and collectives | organizations per 1,000 developers, Q1 2026 | 25% |
| Momentum | Direction of travel | developer count YoY growth, Q1 2026 vs Q1 2025 | 20% |
Each component is min-max scaled to 0 to 100 across the index universe, which is the 70 economies reporting at least 250,000 developers in Q1 2026 with complete data across all four files. The composite runs 0 to 100. A new reading is computed each quarter as GitHub publishes.
Developer Capacity Index, top 25 economies, Q1 2026
| Rank | Economy | DCI | Scale | Intensity | Org depth | Momentum |
|---|---|---|---|---|---|---|
| 1 | South Korea | 72.3 | 52.0 | 97.7 | 100.0 | 36.5 |
| 2 | United States | 61.5 | 100.0 | 53.7 | 48.9 | 29.4 |
| 3 | Switzerland | 60.1 | 23.3 | 100.0 | 81.6 | 38.4 |
| 4 | Germany | 57.3 | 62.6 | 68.8 | 53.3 | 40.1 |
| 5 | Netherlands | 57.3 | 44.5 | 69.4 | 66.2 | 50.3 |
| 6 | France | 55.7 | 57.1 | 76.7 | 50.1 | 34.2 |
| 7 | Japan | 54.7 | 63.3 | 59.2 | 43.8 | 50.0 |
| 8 | Canada | 54.6 | 56.9 | 64.2 | 64.0 | 27.5 |
| 9 | Poland | 54.3 | 43.4 | 87.6 | 48.7 | 36.0 |
| 10 | India | 54.2 | 96.0 | 50.3 | 8.2 | 54.1 |
| 11 | Sri Lanka | 53.0 | 15.8 | 98.5 | 60.9 | 42.0 |
| 12 | Norway | 52.2 | 14.6 | 92.1 | 72.6 | 33.0 |
| 13 | Brazil | 51.9 | 72.0 | 48.7 | 32.7 | 49.7 |
| 14 | Spain | 51.7 | 49.6 | 70.5 | 43.0 | 42.3 |
| 15 | Singapore | 51.1 | 44.4 | 58.5 | 44.2 | 60.6 |
| 16 | United Kingdom | 50.8 | 63.7 | 57.9 | 46.6 | 27.9 |
| 17 | Vietnam | 50.5 | 50.8 | 52.5 | 50.5 | 47.6 |
| 18 | Israel | 49.6 | 27.1 | 83.8 | 51.9 | 37.8 |
| 19 | Czechia | 49.1 | 18.1 | 87.3 | 55.6 | 39.7 |
| 20 | Sweden | 48.9 | 29.0 | 73.7 | 64.0 | 28.9 |
| 21 | Finland | 48.0 | 15.7 | 78.4 | 62.1 | 40.9 |
| 22 | Denmark | 47.0 | 15.9 | 70.6 | 77.3 | 26.2 |
| 23 | Austria | 46.8 | 18.3 | 71.1 | 58.4 | 44.6 |
| 24 | Portugal | 46.6 | 22.2 | 69.3 | 56.0 | 43.1 |
| 25 | Kenya | 46.4 | 20.4 | 73.1 | 33.7 | 67.9 |
Source: Axis Intelligence Research, Developer Capacity Index, computed on GitHub Innovation Graph Q1 2026. All 70 economy scores ship in the dataset below.
What the index says
South Korea leads because it is the only economy that is simultaneously large, intense and organized. Its 3.15 million developers pushed 9.44 million times in Q1 2026, three pushes per account, and it carries 81.9 organizations per 1,000 developers, the highest of any economy above one million accounts. That is a signature of employed, team-based development rather than individual experimentation.
The United States holds second place entirely on scale. Strip the scale component out and it drops behind fourteen economies, because its push intensity (1.77) and organizational depth (52.2 per 1,000) are mid-pack.
India’s profile is the most lopsided in the index: 96.0 on scale, 54.1 on momentum, and 8.2 on organizational depth. India has 764,137 organizations against 26.7 million developers, or 28.6 per 1,000, less than half the Korean rate. According to Axis Intelligence Research, India is adding developers roughly three times faster than it is adding the organizational containers those developers work inside, which is the specific bottleneck to watch as the crossover with the United States approaches.
Kenya at rank 25, with the sixth-highest momentum score in the whole index, is the entry most likely to be missing from every other ranking of developer ecosystems.
Sarah Mitchell: Organizational depth is the component people will argue with, and it is the one worth keeping. An account that never joins an organization is a person learning or tinkering. An organization is a payroll, a client, a university lab, or a maintainer collective. The ratio is the closest public proxy for the transition from individual to institutional software work, and it is the dimension where the developer map looks least like the population map.
Where per-capita rankings actually come from
GitHub published its own per-capita view in a supplementary dataset prepared for the Global Digital Collaboration Conference in Geneva in September 2026, covering 43 economies. It corrects for VPN distortion by using language-model classification of profile locations to reweight IP-based push counts, then divides by World Bank working-age population.
| Economy | Git pushes per 1,000 people aged 15 to 64 |
|---|---|
| Switzerland | 1,635.0 |
| Estonia | 1,581.5 |
| Norway | 1,524.9 |
| Singapore | 1,513.0 |
| Israel | 1,365.2 |
| Iceland | 1,332.9 |
| Netherlands | 1,331.9 |
| Finland | 1,292.0 |
| Sweden | 1,286.3 |
| Denmark | 1,221.4 |
| Canada | 1,116.0 |
| South Korea | 991.7 |
| United States | 884.7 |
| Brazil | 370.7 |
| Kenya | 167.0 |
| Egypt | 134.3 |
Source: GitHub Innovation Graph supplementary dataset, profile-weighted git pushes Q3 2025 to Q2 2026 per 1,000 working-age population. Selected economies shown; the full 43 ship in the dataset below.
Axis Intelligence Research declined to extend this per-capita basis across the full 70-economy index. GitHub publishes working-age population for those 43 economies only, and splicing a different population source across the remaining 27 would produce a ranking whose rows are not comparable to each other. That declined calculation is published as a retracted row in the dataset rather than quietly dropped.
How Much of GitHub Is Now Agent-Driven?
This is the question that changes what the platform is, and 2026 is the first year with hard numbers attached to it.
Microsoft disclosed three on July 29, 2026: GitHub Copilot has 50 million users, one in three pull requests on GitHub involves an agent, and GitHub Copilot revenue accelerated more than 60% quarter over quarter after a June shift to usage-based pricing. Chief financial officer Amy Hood separately attributed part of Azure’s 43% growth to stronger-than-expected Copilot consumption following that billing change.
| Signal | Value | As of | Source |
|---|---|---|---|
| GitHub Copilot users | 50,000,000 | Jul 29, 2026 | Microsoft FY26 Q4 call |
| Pull requests involving an agent | 1 in 3 | Jul 29, 2026 | Microsoft FY26 Q4 call |
| Copilot revenue growth, quarter over quarter | Over 60% | Jul 29, 2026 | Microsoft FY26 Q4 call |
| Copilot attach rate on GitHub accounts | 22.2% | Jul 29, 2026 | Axis Intelligence Research |
| New developers using Copilot in week one | Nearly 80% | Aug 31, 2025 | GitHub Octoverse 2025 |
| Open source projects with a Copilot-using maintainer | 50% | Aug 31, 2025 | GitHub Octoverse 2025 |
| Coding agent pull requests, May to Sep 2025 | 1,000,000+ | Sep 30, 2025 | GitHub Octoverse 2025 |
The attach rate is straightforward arithmetic on two figures stated in the same sentence of the same call: 50,000,000 divided by 225,000,000 is 22.2%.
One tempting calculation does not survive scrutiny and is not published here. Multiplying the one-in-three agent share from July 2026 by Octoverse’s 43.2 million monthly merged pull requests from August 2025 would produce a confident-looking monthly count of agent-involved merges. The two figures are eleven months apart, and one counts merged pull requests while the other counts pull requests involving an agent at any stage. The resulting number would be precise and meaningless. Readers who want the trajectory should note instead that coding agents produced over one million pull requests between May and September 2025, and that eight months later a third of all pull requests carried agent involvement.
Sarah Mitchell: One in three is the number to hold onto, because it is a denominator claim, not a usage claim. Copilot seat counts tell you how many people have access. A share of pull requests tells you how much of the actual merge pipeline has changed hands. The interesting follow-on is not whether the share keeps rising. It is what happens to review capacity when agents author faster than humans approve, which is the pattern GitHub’s own security data already shows for open source vulnerability remediation.
Which Programming Languages Are Growing Fastest on GitHub?
Two public sources measure this differently and disagree in useful ways.
Octoverse ranks by monthly distinct contributors and recorded the decade’s biggest ranking change: TypeScript passed both Python and JavaScript in August 2025 with 2,636,006 monthly contributors, up 66.63% year over year, ahead of Python by about 42,000. Python reached roughly 2.6 million contributors (+48.78%) and JavaScript 2.15 million (+24.79%).
The Innovation Graph measures something narrower: the number of developers in each economy who made at least one push to a repository in that language during the quarter. Summing those across all reporting economies produces a different league table.
Language pushers on GitHub, Q1 2026
| Rank | Language | Pushers (Q1 2026) | YoY growth |
|---|---|---|---|
| 1 | HTML | 5,874,332 | +59.8% |
| 2 | JavaScript | 4,950,656 | +52.6% |
| 3 | CSS | 4,693,314 | +50.0% |
| 4 | Python | 3,532,065 | +67.0% |
| 5 | TypeScript | 2,416,042 | +102.7% |
| 6 | Shell | 1,922,859 | +65.5% |
| 7 | Dockerfile | 1,446,737 | +70.3% |
| 8 | Java | 921,623 | +16.7% |
| 9 | C++ | 794,070 | +27.6% |
| 10 | Jupyter Notebook | 770,538 | +20.8% |
| 11 | C | 757,089 | +26.2% |
| 12 | Makefile | 662,535 | +36.6% |
| 13 | PHP | 467,251 | +12.3% |
| 14 | C# | 458,782 | +17.0% |
| 15 | PowerShell | 442,856 | +130.4% |
Source: Axis Intelligence Research aggregation of GitHub Innovation Graph language file, Q1 2026 versus Q1 2025, EU aggregate excluded.
The two methodologies are not reconciled here, and that is deliberate. Octoverse counts monthly contributors in a language across the whole platform; the Innovation Graph counts quarterly pushers per economy and only where an economy clears the reporting threshold. The first shows TypeScript at number one; the second shows it fifth by volume and first by growth, at +102.7%. According to Axis Intelligence Research, both readings support the same conclusion through different arithmetic: TypeScript is the fastest-moving major language on GitHub, and the typed-JavaScript shift that Octoverse called the largest in a decade is still accelerating rather than settling.
The economy-level split is the part nobody publishes. TypeScript’s share of combined TypeScript-plus-JavaScript pushers ranges from 37.5% in Israel, 36.8% in Singapore and 36.5% in Hong Kong down to 29.1% in India, 29.0% in Russia and 27.1% in Indonesia. The typed-language transition is running roughly ten percentage points ahead in small high-income markets compared with the largest emerging developer populations, which is a hiring signal for anyone staffing front-end work across regions.
PowerShell at +130.4% and Dockerfile at +70.3% are the sleeper entries. Neither is a language people choose for pleasure. Both are what teams write when they are wiring automation and sandboxing runtimes, which is consistent with Octoverse’s finding that repositories containing a Dockerfile grew 120% year over year to 1.9 million.
What Are Developers Actually Building? Topic-Level Signals
Repository topics are self-assigned, which makes them a weak population estimate and an unusually good leading indicator. A topic only spreads when developers start describing their work with the same word.
AI-related repository topics by pushers
| Topic | Q1 2024 | Q1 2025 | Q1 2026 |
|---|---|---|---|
| ai | 5,265 | 10,872 | 33,564 |
| llm | 3,217 | 6,672 | 20,918 |
| machine-learning | 11,032 | 11,028 | 20,928 |
| mcp | not reported | 307 | 10,500 |
| ai-agents | not reported | 278 | 10,518 |
| rag | 435 | 1,182 | 6,092 |
| openai | 2,083 | 2,541 | 6,048 |
| agents | 118 | 765 | 3,794 |
| generative-ai | 569 | 893 | 2,316 |
| deep-learning | 6,091 | 5,010 | 7,882 |
| automation | 2,613 | 3,481 | 13,871 |
Source: Axis Intelligence Research aggregation of GitHub Innovation Graph topics file. Values are counts of developers who pushed to at least one repository carrying that topic in the quarter.
According to Axis Intelligence Research, the mcp topic is the fastest standing-start adoption in the public dataset: not reported in Q1 2024, 307 pushers in Q1 2025, 10,500 in Q1 2026. Model Context Protocol reached 37,000 stars within eight months of release according to Octoverse, and the topic data shows the developer base behind it compounding on a comparable curve. The ai-agents topic followed almost the same trajectory to 10,518.
Meanwhile deep-learning sat at 7,882 pushers, below where machine-learning stood two years ago. The vocabulary of the field has moved from model training to model plumbing inside twenty-four months, and the topic labels moved with it. Teams tracking agentic AI deployment against stated enterprise intent will find this dataset a harder-to-game signal than survey responses, because nobody tags a repository mcp for a vendor’s benefit.
Octoverse corroborates the direction at platform scale: more than 1.13 million public repositories now import a large language model SDK, up 178% year over year, with 693,867 of them created in the twelve months to August 2025, and AI-related repositories exceeding 4.3 million.
How Healthy Is Open Source on GitHub?
Activity records and governance records are moving in opposite directions, and the gap is the most under-covered finding in the 2025 data.
On activity, every line went up. Public repositories reached 395 million (+19% year over year), public contributions 1.12 billion (+13%), merged pull requests 518.7 million (+29%), and March 2025 produced 255,000 first-time open source contributors, the largest single month GitHub has recorded.
On governance, the floor did not move. About 63% of public repositories carry a README, holding flat. Only 5.5% ship a contributor guide. Codes of conduct appear in 2%, or one repository in fifty.
| Governance signal | Share of public repositories |
|---|---|
| README present | ~63% |
| Contributor guide present | 5.5% |
| Code of conduct present | 2% |
Source: GitHub Octoverse 2025.
Security is where that gap becomes expensive. Dependabot is now enabled in 2,668,000 projects (+24.27%), average fix time for critical severity vulnerabilities fell from 37 days to 26, and 26% fewer repositories received critical alerts. Against that, Broken Access Control overtook Injection as the most common CodeQL alert, flagged in more than 151,000 repositories, up 172% year over year, which GitHub attributes partly to misconfigured CI/CD permissions and AI-generated scaffolds that omit authorization checks.
According to Axis Intelligence Research, the pattern across these three datasets is consistent: automation has compressed the time to fix a known vulnerability while the volume of newly introduced authorization defects has grown faster than the contributor-guide and review infrastructure that would catch them earlier. Organizations budgeting cybersecurity spending against AI-accelerated development should weight access control review specifically, not application security generally.
Where GitHub Sits in Microsoft’s Business
GitHub does not report separately. What Microsoft discloses on the record still frames the platform’s commercial position.
| Disclosure | Value | As of |
|---|---|---|
| GitHub users | 225,000,000 | Jul 29, 2026 |
| GitHub Copilot users | 50,000,000 | Jul 29, 2026 |
| GitHub Copilot revenue growth, quarter over quarter | Over 60% | Jul 29, 2026 |
| Fortune 500 using GitHub | Over 90% | Jul 29, 2026 |
| Microsoft 365 Copilot paid seats | Over 30,000,000 | Jul 29, 2026 |
| Microsoft annual revenue, FY26 | $331,000,000,000+ | Jun 30, 2026 |
| Azure annual revenue, FY26 | $100,000,000,000+ | Jun 30, 2026 |
Source: Microsoft FY26 Q4 earnings call, July 29, 2026.
The June 2026 move from seat-based to usage-based Copilot pricing is the structural change. Hood told analysts that Intelligent Cloud gross margins were pressured by growing Copilot usage but improved through the quarter once the new model took effect, and that the shift contributed to Azure revenue coming in ahead of expectations. Microsoft has never broken out GitHub Copilot annual recurring revenue on an earnings call, so anyone quoting a dollar figure for it is quoting a model, not a disclosure. This page will not add one. Enterprises benchmarking their own corporate AI spending against vendor-reported growth should treat the 60% quarter-over-quarter acceleration as a consumption signal rather than a seat signal, because the pricing basis changed inside the comparison window.
Methodology
Collection. Every figure on this page was retrieved on September 21, 2026 from a source fetched during production. Four primary sources carry the page:
- GitHub Innovation Graph (github.com/github/innovationgraph), the quarterly CC0 dataset of public GitHub activity by economy, covering Q1 2020 through Q1 2026. Files used:
developers.csv,git_pushes.csv,repositories.csv,organizations.csv,languages.csv,topics.csv, and the GDC26 supplementary per-capita dataset. - GitHub Octoverse 2025, covering September 1, 2024 through August 31, 2025, published October 28, 2025 and updated February 28, 2026.
- Microsoft FY26 Q4 earnings call transcript, July 29, 2026, for all platform-level 2026 counts.
- US Bureau of Labor Statistics Occupational Outlook Handbook and Employment Projections program, 2025 base year, 2025 to 2035 projections, last modified August 27, 2026.
Definitions matter here more than usual. GitHub’s Innovation Graph counts a developer as an account located in an economy by mode daily IP location as of the last day of the quarter, excluding bots and accounts flagged as inauthentic. Octoverse counts a developer as anyone with a GitHub account. Microsoft’s 225 million is a user count on the same account basis. BLS counts employed people under SOC 15-1252. These are four different populations and are never summed or averaged on this page.
Formulas.
- Developer growth, year over year:
(developers Q1 2026 / developers Q1 2025 - 1) x 100 - Pushes per developer:
git pushes Q1 2026 / developers Q1 2026 - Organizations per 1,000 developers:
organizations / developers x 1000 - Copilot attach rate:
50,000,000 / 225,000,000 - US accounts per employed developer:
32,410,529 / 1,717,800 - Developer Capacity Index:
0.30 x scale + 0.25 x intensity + 0.25 x organizational depth + 0.20 x momentum, each component min-max scaled 0 to 100 across the 70-economy universe, with scale computed onlog10(developers) - India and United States crossover: OLS on
ln(developers)over eight quarters ending Q1 2026, solved for equality; sensitivity reported against an OLS fit on absolute counts
Scope. Innovation Graph data covers public activity on GitHub only and reports an economy only where 100 or more unique developers performed the relevant activity in the quarter. Activity is attributed by IP address, so VPN use shifts activity between economies; GitHub’s supplementary profile-weighted dataset exists specifically to correct for this and is used here where per-capita comparisons are made. Octoverse public metrics undercount private activity, which accounted for 81.5% of contributions in 2025. Topic counts reflect self-assigned repository topics. All Axis-calculated values are labelled as such in the dataset, and the one calculation declined during production ships as a retracted row with its reasoning.
Methodology version. The Developer Capacity Index methodology is stated in full above. Any change to components, weights or universe will be published with the reading it first applies to, and prior readings will not be restated.
About This Dataset
github-statistics-2026.csv contains 1,196 rows covering GitHub platform metrics, 70 economies across nine metrics each, 20 programming languages, 15 repository topics across three annual snapshots, the 43-economy per-capita series, and every Axis-calculated figure on this page. Each row carries metric, dimension, value, unit, as_of_date, source_org, source_document, source_url, retrieved_date, is_primary, axis_calculated, method_note and data_type.
License: CC BY 4.0. Free to use, redistribute and build on with attribution.
Cite this page
APA Axis Intelligence Research. (2026). GitHub statistics 2026: users, developers by country, languages and agent adoption. Axis Intelligence. https://axis-intelligence.com/github-statistics/
MLA Axis Intelligence Research. “GitHub Statistics 2026: Users, Developers by Country, Languages and Agent Adoption.” Axis Intelligence, 2026, axis-intelligence.com/github-statistics/.
Chicago Axis Intelligence Research. “GitHub Statistics 2026: Users, Developers by Country, Languages and Agent Adoption.” Axis Intelligence. 2026. https://axis-intelligence.com/github-statistics/.
Frequently Asked Questions
Is the 180 million developer figure still current for GitHub?
No. 180 million comes from Octoverse 2025 and describes the platform as of August 31, 2025. Microsoft stated 225 million GitHub users on July 29, 2026. Any 2026 analysis still quoting 180 million is roughly eleven months and 45 million accounts behind. Octoverse 2026 is expected around GitHub Universe on October 28 and 29, 2026.
Which GitHub number should I use in a report, Octoverse or the Innovation Graph?
Use Octoverse for platform-wide annual totals and language contributor rankings; it covers public and private activity and runs on a September-to-August research year. Use the Innovation Graph for anything country-level or anything needing a recent quarter, since it publishes quarterly and reaches Q1 2026. Do not mix them inside a single comparison: Octoverse counts monthly contributors platform-wide, the Innovation Graph counts quarterly pushers per economy, and the two will not reconcile.
How do I reproduce the Axis Developer Capacity Index myself?
Clone github/innovationgraph, filter developers.csv, git_pushes.csv, organizations.csv to Q1 2026 and developers.csv to Q1 2025, exclude the EU aggregate row, keep economies with at least 250,000 developers and complete data across all four files. Compute log10 of developers, pushes divided by developers, organizations divided by developers times 1,000, and year-over-year developer growth. Min-max scale each to 0 to 100 across that universe and apply weights of 0.30, 0.25, 0.25 and 0.20. The full component scores for all 70 economies ship in the dataset above.
Why does China rank third in developers but near the bottom on push activity?
The Innovation Graph assigns activity to economies by IP address, and GitHub’s own documentation states that VPN use and IP masking distort economy-level metrics. China recorded 11.35 million developer accounts and 2.92 million git pushes in Q1 2026, or 0.26 pushes per account, against a top-20 median of 1.65. Part of that gap is real substitution toward domestic code hosting and part is measurement. Country comparisons involving China should be read with both in mind.
Does “one in three pull requests involves an agent” mean agents write a third of all code?
No. The statement, made by Satya Nadella on July 29, 2026, is about pull requests that involve an agent at some stage, which includes agent-authored drafts, agent code review and agent-assisted edits. It is not a measure of lines of code, and it does not state how many of those pull requests merged. The nearest bounded figure is GitHub’s own: Copilot coding agent created more than one million pull requests between May and September 2025.
Has TypeScript actually overtaken Python on GitHub?
By GitHub’s contributor methodology, yes, as of August 2025, and narrowly: TypeScript led Python by roughly 42,000 monthly contributors, with 2,636,006. By the Innovation Graph’s quarterly pusher methodology, no; Python recorded 3.53 million pushers in Q1 2026 against TypeScript’s 2.42 million. TypeScript leads on growth in both, at +66.63% and +102.7% respectively. The ranking depends entirely on whether you count contributors or pushers, and reports that state it as settled have picked one without saying so.
Are GitHub account totals a reliable estimate of the number of software developers in a country?
No, and the ratio is large. US GitHub developer accounts stood at 32,410,529 in Q1 2026 while BLS counted 1,717,800 employed US software developers in 2025, a ratio of 18.9 to one. Accounts include students, hobbyists, researchers, duplicate and dormant registrations, and people who write code outside the BLS occupation classification. Use account data for platform reach and labour statistics for workforce size.
What does the mcp repository topic tell us that survey data does not?
It tells you what developers are already building rather than what their employers say they intend to build. The topic went from not reported in Q1 2024 to 307 pushers in Q1 2025 to 10,500 in Q1 2026. Repository topics are self-assigned at the point of work, carry no vendor incentive, and are published quarterly under CC0, which makes them the cheapest available leading indicator for protocol and tooling adoption.
Which economies are growing fastest on GitHub right now?
Among economies with at least 250,000 developer accounts, the fastest year-over-year growth in Q1 2026 came from Iraq (+63.8%), Uzbekistan (+53.3%), Ethiopia (+52.9%), Algeria (+49.1%), Saudi Arabia (+47.2%), Ghana (+47.1%) and Kenya (+46.3%). Among the twenty largest economies, Pakistan (+45.3%), the Philippines (+41.3%) and India (+38.8%) lead. Every one of those exceeds the United States at +25.4%.