Lifestyle

GitHub and Yale survey: 80% of developers surveyed want energy-efficient coding tools but lack ways to measure impact

On September 23, 2026, GitHub published a survey of 1,039 US GitHub users run with the Yale Program on Climate Change Communication. Most respondents want tools to make software use less energy but lack ways to measure it. Here are the key figures, the sample's limits and what it means for everyday users and development teams.

About 6 min read

GitHub and Yale survey: 80% of developers surveyed want energy-efficient coding tools but lack ways to measure impact
Image: Mokaair (Original editorial artwork)

What happened

GitHub, the widely used platform where developers store and share code, published a research post on its official blog. The post was written by Paull Young, released on September 23, 2026 and updated on September 24. According to GitHub, the survey was run jointly by GitHub and the Yale Program on Climate Change Communication. It collected 1,039 responses from monthly active GitHub users in the US and asked about climate change, AI, software efficiency and the responsibilities of organizations across the tech industry.

GitHub's core conclusion is that developers know efficient software matters. However, many lack a clear way to find waste, measure an improvement and make the case for fixing it.

GitHub and Yale survey: 80% of developers surveyed want energy-efficient coding tools but lack ways to measure impact
Mokaair editorial verification flow · Image: Mokaair (Original editorial artwork)
Read the full description

Sources are collected, independently checked, then reviewed by Jev.

Key survey figures

According to GitHub, 79% of respondents are worried about global warming. 71% are concerned about the environmental impact of AI systems, including their energy and water use and carbon emissions. 75% say it is important for their employer to actively reduce its environmental impact. When asked what they want, 80% are interested in energy-efficient coding tools and 78% want to learn best practices for reducing software's environmental footprint. 74% want to measure the environmental impact of their software or development process, and 70% are interested in contributing to open source projects focused on sustainability.

GitHub also compared some answers with results for US adults from "Climate Change in the American Mind", Yale's nationally representative survey. The GitHub users surveyed expressed more concern about climate change than US adults overall.

Figures as published by GitHub. GitHub notes the differences reflect both different populations and different survey designs.
QuestionGitHub respondentsUS adults
Think global warming is happening86%68%
Say it is at least somewhat important to them personally82%65%
Think it will harm them at least a moderate amount68%45%
Expect it to harm future generations at least a moderate amount82%68%
Worried about global warming79%66%

Where the gap lies: willingness, but no practical path

According to GitHub, only 10% of respondents believe the way they write software does a lot to reduce their personal environmental impact. Another 28% say it has a moderate effect, and 63% say the effect is small. In their written comments, some respondents asked for ways to estimate the footprint of repositories (the storage space for a project's code) and CI/CD workflows (automated pipelines that build and test code). Others wanted to find unnecessary runs of GitHub Actions, GitHub's automation service, or to compare AI use with other sources of computing demand. Several respondents also warned against making environmental claims without evidence.

GitHub specifically notes that faster code can reduce resource use, but running time alone does not prove lower energy use or emissions. Hardware, workload, location, time and the source of electricity all affect the result.

GitHub's recommended approach and AI agent tools

GitHub recommends looking for measurable waste in four areas: - Code, such as repeated computation and inefficient algorithms. - Data, such as fetching more than needed and missing caching. - Network and I/O (input and output), such as duplicate requests and oversized payloads. - Front end, such as unnecessary rendering and loading off-screen assets too early. Execution time, CPU use, memory allocation and network transfer size can all serve as proxy metrics. These are indirect measures of how much computing a program needs. GitHub cautions that each has limits, so teams should state what they measured and what they did not.

GitHub also introduced its open source Daily Efficiency Improver workflow, part of GitHub Agentic Workflows. These workflows use an AI agent, an AI tool that carries out tasks on its own. According to GitHub, the workflow reviews a repository for efficiency opportunities and runs the repository's tests. It can then open draft pull requests, which are proposed code changes for people to review, with the evidence and trade-offs attached. It does not merge changes itself. Before enabling it, GitHub recommends reviewing its permissions, configuration, model use, run frequency and likely compute cost. GitHub also suggests starting in a test repository or running the workflow manually. GitHub says humans still decide whether the evidence is sound and whether a change is adopted.

What this means in practice

  • For everyday users: the survey shows that some developers care about the energy use of software and AI. However, it is an opinion survey and does not mean any service has actually become more energy-efficient.
  • For development teams: the core of GitHub's advice is "measure first, then claim". Back efficiency improvements with concrete data and reproducible tests rather than simply calling a change "greener".
  • For teams considering AI agent tools: GitHub advises checking permissions and likely compute cost before enabling a scheduled workflow. It also advises treating every recommendation as a hypothesis until benchmarks and tests support it, with maintainers deciding what ships.

Frequently asked questions

Does this survey represent the views of all engineers?

No. According to GitHub, the respondents were US users who had agreed to receive GitHub marketing emails. They were not randomly selected, so the results describe only the respondents' views.

Does the survey prove that AI uses a lot of energy?

No. GitHub explicitly states that the survey did not measure the environmental footprint of AI or any software system. It only reflects respondents' level of concern.

Does faster code mean lower energy use?

Not necessarily. GitHub notes that running time alone cannot prove lower energy use or emissions, since hardware, workload, location, time and electricity source all affect the result.

Will Daily Efficiency Improver automatically change my code?

According to GitHub, it can open draft pull requests for maintainers to review, but it does not merge changes itself. People decide whether to adopt them.

Who produced these figures?

The figures come from a survey GitHub ran with the Yale Program on Climate Change Communication and published on the official GitHub Blog. They are reported here as GitHub published them.

Browse the latest news in this topic

More on this topic

Latest travel guides

Sources

Lifestyle