Discussing GitHub Copilot, Microsoft reported that developers using Copilot "are 55% more productive" on certain tasks and that "about 40% of the code they check in is now AI-generated and unmodified." This statistic is limited to Copilot users studied and does not indicate that 40% of all new source code in the U.S. is AI-generated.

A study published in the journal *Science* earlier this month looked at GitHub Python functions and found that about 29% in the U.S. are now AI-written, with lower percentages in other geographies. Microsoft CEO Satya Nadella said in April 2025 that AI was generating about 30% of code at the software giant. Salesforce has given out a similar figure, suggesting that for some large U.S. companies, roughly a third of new code is already AI-assisted.

Business Insider reports that Google executives said in 2026 that "three-quarters of new code created inside Google is now generated by AI" and reviewed by human engineers. The article notes that as of October 2024, around a quarter of the company's code was AI-generated, rising to 50% last fall, before reaching 75%. It also mentions that Snap disclosed that at least 65% of new code under its new operating model is generated by AI, and that Microsoft executives have cited 20–30% AI-written code for some projects.

84% of respondents are using or planning to use AI tools in their development process, an increase over last year (76%). This year we can see 51% of professional developers use AI tools daily. While the survey measures tool usage rather than the percentage of code written by AI, it indicates that AI coding assistants have become part of everyday practice for a majority of developers.

At Meta’s LlamaCon conference, Microsoft CEO Satya Nadella said that "20% to 30% of code inside the company's repositories" was written by software, meaning AI. Nadella characterized the figure as applying to code in Microsoft's repositories at that time, with stronger AI usage in languages like Python and less in C++. TechCrunch adds that Google CEO Sundar Pichai recently said AI was generating more than 30% of Google's code, while cautioning that these corporate figures depend on how AI-generated code is defined and measured.

AI coding tool adoption reached 84–91% across four major surveys in 2025–2026, yet trust in AI accuracy dropped to 29% among Stack Overflow respondents. The broadest count is Stack Overflow's: 84% of all developers use or plan to use AI tools in their development process, up from 76% in 2024. One additional data point from GitHub: 80% of new developers on GitHub use Copilot within their first week, according to the GitHub Octoverse 2025 report.

METR describes experiments with open-source developers using AI coding assistants like Claude Code and Codex, but focuses on productivity effects rather than the overall share of AI-written code. Their 2025 study found that AI tools initially caused tasks to take 19% longer on average, with a later follow-up suggesting a possible speedup among some participants. The blog notes an increase in the use of "agentic" coding tools throughout 2025, indicating adoption growth but providing no estimate that more than 30% of all newly written U.S. code is AI-produced.

In describing enterprise use, the market report cites a concrete example: "NatWest reports that **12,000 engineers now let AI write more than 35% of their production code**, while agentic workflows deliver tenfold productivity gains in its financial crime units." Elsewhere it notes that the AI code tools market is growing rapidly, with North America holding a major share, but it does not give a single global percentage for all new code that is AI-assisted. The NatWest example illustrates that at least some large organizations report AI involvement in more than a third of their production code.

One number circulates constantly in press coverage: 41% of code is now AI-generated. It comes from Emad Mostaque, who took GitHub’s figure about the share of code accepted by Copilot users and extrapolated it into a claim about all code everywhere. The original figure applied only to developers already using Copilot, a fraction of GitHub’s user base at the time; the extrapolation doesn’t hold. The more defensible numbers: DX’s measurement across 4.2 million developers puts AI-generated production code at 26.9%. A study published in *Science* found roughly 30% of Python functions from U.S. contributors on GitHub were AI-generated by late 2024.

The 2024 Stack Overflow Developer Survey reports that a majority of professional developers have tried or regularly use generative AI tools such as ChatGPT and GitHub Copilot, but the survey does not quantify the exact percentage of their code written by AI. Instead, it measures frequency of tool usage and attitudes toward AI-assisted development. The results show rapid adoption of AI coding aids but do not support a specific figure like "more than 30%" of all newly written U.S. code.

A 2025 industry analysis summarizing multiple sources notes that "GitHub's enterprise data from Copilot usage across large deployments supports the industry average range of 15-25% AI-assisted lines." It clarifies that this 15–25% range is an estimated share of AI-assisted lines within organizations that have deployed such tools, not an estimate that 15–25% of *all* code written in a country is AI-generated.

GitClear’s 2025 report, based on analysis of commits in public and private repositories, writes that AI coding assistants have driven "a roughly 4x increase in AI-generated code between 2022 and 2024" and cites Stack Overflow data that "63% of professional developers said they currently use AI in their development process." The report focuses on growth and quality trends and does not provide a statistic that more than 30% of all newly created code in a specific country (such as the U.S.) is AI-generated.

Discussing adoption and impact in the U.S. market, the article states: "With **91% developer adoption** and **42% AI-assisted code**, the focus now shifts from whether to adopt AI coding tools to how to manage and measure their impact." It frames these numbers as 2026 US market data, describing that developers use a median of 3.1 AI coding tools and that GitHub Copilot and Claude Code are leading tools. The figures are presented as estimates for the share of code that involves AI assistance rather than fully autonomous generation.

Quick answer: As of 2026, 84% of developers use or plan to use AI coding tools (Stack Overflow 2025, n=49,000+), but only 29% trust the output—down from 40% in 2024. In the US and Canada, adoption reached 24%, indicating regional differences in how widely AI tools are integrated into development workflows. These figures reflect adoption and trust rather than a direct percentage of code written by AI.

In a 2024 blog post describing internal experiments, an engineering leader writes that in one pilot, "commit-based metrics revealed that the engineering team's self-reported AI usage was 80% while the tracked metric showed 65%" AI-written code for that team. The post emphasizes that this result comes from a specific organization and cautions that such percentages depend heavily on context, tooling, and measurement method, and thus cannot be generalized to all software development or to an entire country.

According to recent global estimates, 41% of all code is now AI-generated, with 76% of professional developers either using (62%) or planning to use (14%) AI coding tools. As of early 2026, the share of AI-generated code has surged to near 50%, with adoption curves steepening faster than initial projections. However, the article later notes that Google CEO Sundar Pichai said around 25% of Google’s code is AI-assisted, and Microsoft CEO Satya Nadella has said AI now writes about 20–30% of the code in some of their projects, indicating variation across organizations.

GitHub has reported in public presentations and blog posts that, for developers who use GitHub Copilot, around 40–50% of the code they ultimately accept in the editor is suggested by the AI assistant. These figures apply only to active Copilot users and specific languages and do not represent the share of all newly written code in the United States, but they are often cited when discussing how much code in individual workflows is AI-generated.

A 2023 Hacker News discussion examines the statement that "today, almost 50% of code is written by AI" and traces it back to Microsoft’s comment that "40% of the code [Copilot users] are checking in is now AI-generated and unmodified." Commenters point out that this original figure applied only to developers using Copilot in the cited study, not to all code written worldwide or in any specific country, and that extrapolating it to say half of all code is AI-written is misleading.