This report draws on insights from the JetBrains Developer Ecosystem Survey 2025 to explore how C++ developers are working today – from the standards they adopt and the tools they rely on to their testing approaches, build systems, and the growing role of AI. Based on responses from 1,800 C++ developers from 22 countries, The State of C++ 2025 shows how C++ is adapting to modern development trends while staying true to its performance-oriented roots.
From embedded systems to large-scale applications, C++ continues to demonstrate its versatility. We hope this report helps you see how the language and its ecosystem are evolving and where your own work fits into that story.

| 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | |
|---|---|---|---|---|---|---|---|
| – | – | – | – | – | – | 7% | C++26 |
| – | – | – | – | 10% | 15% | 21% | C++23 |
| – | 12% | 18% | 23% | 29% | 36% | 34% | C++20 |
| 26% | 41% | 42% | 41% | 43% | 45% | 43% | C++17 |
| 33% | 39% | 30% | 25% | 21% | 19% | 16% | C++14 |
| 63% | 46% | 40% | 31% | 27% | 26% | 24% | C++11 |
| 18% | 13% | 12% | 8% | 8% | 7% | 9% | C++98 / C++03 |
| – | – | 11% | 14% | 13% | 8% | 12% | I'm not sure |
of C++ developers were early adopters of the C++26 standard in 2025. By contrast, most upgrade plans point to C++20 or C++23, indicating the slow pace of C++26 adoption.
C++14 is losing the most users, with 60% planning to switch to newer standards.
remains the most common standard (43%), but C++20 (34%) and C++23 (21%) are gaining traction.
More than two-thirds of C++ developers consider their current standard sufficient. 28% cite dependency constraints as their reason not to upgrade, and 17% lack the resources required for migration.
Note: Despite all the measures we've taken to secure a representative pool of respondents, the results may be slightly skewed toward users of JetBrains products, as they may have been more likely to participate in the survey.
VS Code (33%), CLion (23%), and Visual Studio (21%) lead the field, together accounting for more than three-quarters of C++ developers. The rest of the market is split among niche tools like Vim, Qt Creator, and Rider, each with only a few percent.
| 2021 | 2022 | 2023 | 2024 | 2025 | |
|---|---|---|---|---|---|
| 25% | 26% | 23% | 27% | 26% | ReSharper C++ |
| 22% | 27% | 22% | 17% | 14% | Visual Assist |
| 11% | 11% | 8% | 11% | 10% | ReSharper |
| 6% | 5% | 8% | 5% | 9% | CodeMaid |
| 3% | 5% | 4% | 4% | 3% | AWS Toolkit for Visual Studio |
| 3% | 2% | 3% | 5% | 5% | CodeRush |
| 1% | 1% | 1% | 2% | 2% | Roslynator |
| 6% | 8% | 5% | 4% | 3% | Other |
| 48% | 42% | 49% | 52% | 49% | None |
ReSharper C++ (26%) and Visual Assist (14%) are the top plugins, while nearly half of Visual Studio users report not using any plugins at all.
GoogleTest is the most widely used framework, used by 31% of C++ developers, yet more than one-third skip unit tests entirely.
32%
30%
Boost
28%
24%
Qt
24%
26%
OpenCV
20%
22%
OpenSSL
17%
15%
zlib
15%
15%
libcurl
15%
9%
TensorFlow
Boost (32%), Qt (28%), and OpenCV (24%) remain the leading libraries, while TensorFlow adoption has grown to 15%.
Most C++ developers manage libraries using system package managers (36%) or include source code directly (35%). Dedicated tools like vcpkg (11%) or Conan (6%) see comparatively limited use.
CMake dominates C++ build systems at 59%, ahead of Visual Studio projects (34%) and Makefiles (27%).
GCC (70%), Clang (45%), and MSVC (27%) dominate.
Python stands out as the most common companion to C++ projects (51%), with C (30%) and shell scripting languages (21%) also widely used.
ClangFormat (29%), IDE features (28%), and Clang-Tidy (27%) are nearly tied as the leading tools for code quality. Still, more than a quarter of developers say they don’t use any such tools.
The Google C++ Style Guide leads at 31%, with Clang-Tidy (26%) and ClangFormat (25%) following close behind. One-third of developers report not following any style guide.
44%
50%
Very likely
21%
21%
Somewhat likely
11%
10%
Not sure
7%
4%
Somewhat unlikely
8%
5%
Very unlikely
8%
11%
I already use AI coding agents
42%
43%
Very likely
24%
21%
Somewhat likely
14%
15%
Not sure
6%
5%
Somewhat unlikely
7%
5%
Very unlikely
7%
10%
My company is already using AI coding agents
54%
52%
Data privacy and security concerns
48%
29%
Concerns about intellectual property rights over code generated by such tools
19%
19%
Difficulty in incorporating such tools into existing workflows and systems
18%
13%
Distrust of new technologies
16%
19%
Processes and decision-making for adopting new technologies are slow and conservative
16%
14%
Coders don’t see value in such tools or services
15%
13%
We have no resources for investigating the benefits and risks of adopting such tools
44% of C++ developers are very likely to try AI coding tools in the next year, compared to 50% of developers who use other languages. Just 8% of C developers are already using AI, with security and IP concerns cited as the main blockers to adoption, followed by slow internal processes and lack of awareness.
45%
55%
Third-party cloud-based AI tools or services
27%
18%
Organization’s own cloud-based AI tools or services
21%
16%
Organization’s own local, non-cloud-based AI tools or services
18%
13%
Third-party cloud-based AI tools or services managed and hosted by my company
14%
11%
Third-party locally installed, non-cloud-based AI tools or services
6%
9%
I don’t know
1%
1%
Other
12%
11%
None
Among C++ developers, cloud-based third-party AI tools lead with 45% usage. Local or organization-hosted solutions rank second, with 27% adoption.
8%
8%
Less than 1 hour
23%
20%
From 1 to less than 2 hours
28%
30%
From 2 to less than 4 hours
19%
19%
From 4 to less than 8 hours
16%
19%
8 hours or more
6%
4%
I don’t save any time
Most C++ developers report saving 2–8 hours weekly thanks to AI tools.
30%
23%
Quality of generated code
17%
10%
Negative effect on my coding and development skills
13%
19%
Limited understanding of complex code and logic by AI tools
11%
10%
Lack of context awareness
10%
13%
Privacy and security
7%
5%
Costs
5%
10%
Job security and future job prospects
The most common concerns include code quality, skill degradation, and AI’s limited understanding of complex code.
| Not important at all | Slightly important | Moderately important | Very important | |
|---|---|---|---|---|
| 7% | 29% | 30% | 33% | Availability of different AI models / LLMs |
| 13% | 24% | 38% | 25% | Availability of local AI models / LLMs that do not require internet access |
| 2% | 8% | 40% | 51% | Availability of the best AI models / LLMs for coding |
| 5% | 20% | 29% | 46% | Easy-to-use user interface |
| 5% | 26% | 42% | 27% | Frequency of code completion suggestions |
| 1% | 15% | 41% | 43% | Predictability of features’ behavior |
| 4% | 17% | 34% | 45% | Price |
| 1% | 6% | 27% | 66% | Quality of answers in chat |
| 2% | 13% | 25% | 61% | Quality of code completion suggestions |
| 1% | 2% | 12% | 85% | Quality of generated code |
| 1% | 6% | 19% | 73% | Quality of the AI features’ codebase context awareness |
| 5% | 10% | 21% | 63% | Security and privacy |
| 3% | 6% | 30% | 60% | Smooth integration with my IDE or code editor |
| 1% | 19% | 50% | 30% | Speed of code generation (apart from code completion) |
| 2% | 17% | 42% | 38% | Speed of generating code completion suggestions |
| 1% | 27% | 45% | 27% | Speed of response generation in chat |
When using AI, C++ developers value the quality of generated code, IDE integration, model availability, and predictability of behavior. Speed and privacy also rank highly.