loading
loading
With fall upon us, it’s time to pumpkin spice up your life with the tasty results of the 2026 Developer Survey. Over seven weeks, more than 30,000 people took the survey and answered questions about their work, technologies, learning strategies, and, yes, AI use. We diligently worked in our data mines to harvest insights that you can use to build better software and engineering teams. We’ll be releasing the full data set SoonTM so you can find data to back your own hot takes. Below, we’ll highlight some of the results we found interesting. If you want to skip ahead to the results themselves, check out the Survey overview then dive into the results. This is the 16th Developer Survey, which means it’s old enough to drive in the US. Hand the keys over to us, and let’s get rolling! Work Work for developers has certainly changed, and only some of that is due to AI. Only a small fraction (18%) of devs go into offices every day to work. Those that do probably aren’t going into offices with the ping pong tables, endless snack cabinets, and tactical bean bag chairs. AI and layoffs have changed the calculus between software developers and employers. Before 2020, you could walk to the corner store and trip over a job offer. Now LinkedIn profiles declare, “Open to Work!” for months on end. But if companies won’t hire you, maybe you can hire yourself. While about the same percentage of respondents said they were freelancers, contractors, or self-employed, those that said they worked in an organization of one jumped from 4% to 10%. We’ve heard anecdotally about how some folks are using AI agents to soloprenuer their way to a viable company, so this may be the data to back it up. See more results about how developers work today. Technology For plenty of you, the technology rankings are the meat of the survey. There’s not a ton of movement among the leaders—the top ten is mostly the same, though SQL passed HTML/CSS to take the number two spot. Meanwhile, Rust increased its standings by moving up to the 12th most used language. Perhaps that White House memo about memory safety made an impact? For AI fans, the top model providers, Anthropic and OpenAI, are neck and neck, with Google a distant third. Similar levels of usage don’t translate to similar levels of admiration: more respondents want to keep using Anthropic models than OpenAI models, though not by much. AI There’s no question that AI has become prevalent in the software industry—even if that prevalence has some folks feeling frustrated. 73% of respondents who indicate using AI coding assistants/agents use them daily, and a third of daily users using any AI tools use AI four or more hours per day. With that increased use has come greater positive sentiment—62% of all respondents view AI at least somewhat positively, while that jumps to 71% for daily users . Many of those AI users live in coding agents (66%) and chatbots (63%). No surprise; developers use the tools that are best for developing. The agents of choice are Claude Code (66%) and GitHub Copilot (59%). When they develop with AI, it’s primarily generating code in familiar areas (67%) and debugging (61%), not deploying, operating, or troubleshooting production systems (20%). AI gets to muck about with familiar code, but once that code reaches prod server, it gets shown the door. Community AI has definitely changed how people find answers to their questions: 83% still ask a search engine for answers, but 70% hit up their AI. Those search engines used to bring folks to a Stack Overflow question, but AI overviews and chatbots don’t. They still use our answers, but they don’t always give the person who answered the credit that they deserve. People are visiting less for simple questions (64%), but many are visiting more just to check the source of information for those AI answers (16%). While that’s less than last year, it still indicates that there are people increasing how much they visit even when using AI. When developers are looking to learn something new, what skills are they learning? No surprise, but a large majority (79%) want to upskill their AI game. It’s the new hotness. Everybody is being asked to get onboard or get left behind, so these are skills that need attention. But that doesn’t mean your standard programming skills gather rust and dust: 52% of people are still learning new coding skills. Knowledge, trust, and context Context is the big buzzword in agentic engineering right now, and we wanted to know what that means for everyone. We’ve seen that good context is trusted context, and to trust what the AI gives requires source attribution (93%). The best sources of trusted context are project goals and requirements (85%) and code and docs (73%). The places to look for specific answers are the people you work with (72%), the code (63%), and documentation (60%). That explains why senior devs spend half their day answering DMs. Developers and their employers have started to pay attention to token costs as AI use rises. We’ve seen model routing and usage limits emerge as solutions. But developers care more about quality (75%) over cost (38%) when it comes to shaping prompts and picking models. Their organizations (and their CFOs no doubt) have placed some sort of cost-control measures in place: 38% set limits on usage, 28% reviewed spending data, 27% use different models for different work, and 25% set limits on who gets to use AI. But some orgs take the developer’s side: 21% said their organizations are spending more. Check the full results yourself This is just a little sample of what the survey results contain, so please do go over to the survey site and browse the full details yourself. We’re taking a more editorial approach this year, doing what we can to highlight more of the data points that we think show interesting trends in the industry. If you want to skip over the chat and hit the numbers directly, feel free. In a few weeks, we’ll have the full data set available for download so you can comb through the numbers yourself. If you find anything interesting or have alternate, complimentary, or disagreeable takes off of what you see, let us know. We’re always interested in what the community thinks. If you want to check how we’re looking to solve the context and knowledge issues found above, check out Stack Internal.
document fetched 2026-10-24 10:03 · canonical https://stackoverflow.blog/2026/10/06/the-results-of-the-2026-developer-survey-are-here