Claude Code Overtakes GitHub Copilot: What JetBrains' Survey of 15,000 Developers Says About AI Coding Agents in 2026

AI coding agents are no longer an experiment. According to JetBrains' Developer Ecosystem Survey 2026, 90% of professional developers now use AI coding agents at work at least weekly. 68% use them daily. The survey ran from May to July 2026 and collected responses from more than 15,000 professional developers worldwide, in its tenth year running. But the headline number is not the interesting part. The interesting part is who won and who lost. In under a year, the market leader changed, and the tool that defined AI-assisted coding is now in decline.

Claude Code Went From 18% to 39% in Six Months

Claude Code is now the most widely adopted AI coding tool at work. Around 39% of professional developers worldwide used it in May to July 2026, up from 18% in January 2026. In the United States the number is 47%, meaning almost half of US developers use it at work.

One stat stands out. Claude Code is the single most used AI coding tool for 31% of developers. JetBrains frames this as an almost 80% conversion rate: most developers who try it at work end up making it their main tool. That is unusually sticky for a tool in a crowded market. It is now used roughly twice as often as GitHub Copilot, the tool that made AI-assisted coding mainstream back in 2023.

GitHub Copilot Lost the Lead

Copilot's adoption dropped from 29% a year ago to 21% in May to July 2026. It is no longer the leader. The twist is that everyone still knows it. Copilot's awareness sits at 79% worldwide, and 86% to 90% in Europe, the UK, and the US. That gap is the story: developers know Copilot but increasingly choose something else. Awareness without adoption usually means a product that people tried and moved on from.

One healthy sign for GitHub: 39% of Copilot users run it inside JetBrains IDEs, so the tool still lives across many surfaces. But being everywhere is not the same as being first choice.

Codex Is the Fastest Riser

OpenAI's Codex grew roughly 5x in six months, from 3% adoption in January 2026 to 16% in May to July 2026. Its awareness jump may matter more. In January, only 27% of developers had even heard of Codex. By May to July, that number reached 65%. OpenAI turned an unknown product into a mainstream option inside half a year. If the adoption curve follows awareness the way Claude Code's did, Codex is the one to watch next.

Cursor Slipped, Especially in China

Cursor's adoption fell from 18% to 12%, even though its awareness rose from 69% to 75%. More people know it, fewer use it. The sharpest drop is in China, where Cursor went from 28% adoption in January to 16% in May to July. Local alternatives and pricing likely play a role there, though the survey data does not break down the reasons.

The Open-Source Entry Worth Noting

OpenCode, an open-source coding agent, reached 7% adoption and 42% awareness. No big company stands behind it. For a market dominated by Anthropic, OpenAI, GitHub, and Google budgets, an independent open-source tool reaching 42% mind share is a signal. Developers care about the agent concept itself, and the brand on the box comes second.

Google's Antigravity held stable at 6% adoption while its awareness climbed from 29% to 47%. In India it is tied with Cursor as the third most popular tool, at 15% adoption. JetBrains' own AI tools sit at around 9%.

What This Means If You Pick Tools for a Team

Three practical takeaways from the data:

• The market churns fast. The leader changed twice in two years. Avoid deep integration with any single agent's proprietary features. Skills like prompt discipline, review discipline, and good task breakdowns transfer across tools. Vendor-specific workflows do not.
• Daily use means the agent is infrastructure now. When 68% of developers use agents daily, an outage or pricing change is a team productivity event, not a personal annoyance. Budget and plan for it the way you plan for CI.
• Conversion rate beats adoption rate. Copilot has the awareness, Claude Code has the conversion. When you pilot a tool, measure how many engineers keep using it after a month, not how many tried it once.

Read the Numbers With Care

The survey is statistically reweighted to represent the global developer population by region, employment status, programming language, and familiarity with JetBrains products. It was localized into eight languages. Still, it is one vendor's survey, adoption here means self-reported use at work, and respondents familiar with JetBrains products may skew slightly toward IDE-heavy workflows. The direction of the trends is consistent across independent coverage from heise online and others, but treat exact percentages as estimates, not measurements.

The bigger picture is simple. In January 2026, autocomplete-era tools still led. By mid-2026, autonomous agents took over, and the developers who use them daily are now the majority. If your team still treats AI coding tools as optional plugins, the survey suggests your competitors probably do not.