The Solidity Developer Survey 2025 results are out! See the interactive report
{ skip to content }

Solidity Developer Survey 2025

Key findings

  • 70% of respondents are smart contract developers, with India, Nigeria, and the US as the top countries
  • Foundry is the dominant framework at 57%, up from 51% in 2024. Truffle is down to a single user.
  • Stack-too-deep remains the #1 pain point (47%), with experts reporting it more than beginners (65% vs 25%)
  • 88% use AI tools at least monthly, but adoption outpaces trust: 45% express distrust in AI output
  • Only 30% of respondents are familiar with Core Solidity. Among those who are, better error handling and delegatecall replacement are the most wanted features. Learn more about Core Solidity.
  • DevEx is improving: 73% report improvement (up from 67% in 2024)

Survey Overview

The Solidity Developer Survey 2025 covers the 2025 development year and was conducted in February-March 2026. It received 640 fully completed responses and 1,095 usable responses after removing empty submissions. Respondents came from 87 countries.

Charts use per-question N: every response that answered a given question is included regardless of overall survey completion. All charts are interactive: hover for details.

This report presents the data as collected, without editorial interpretation. For our takeaways and planned actions, see the accompanying blog post.

Download the raw data (CSV) | Previous surveys: 2024, 2023

Respondents per section

n = 1,095 usable responses

Respondent counts decrease through the survey as some participants drop off before completing all pages.

Demographics

Where do you live?

n = 999

India, Nigeria, and the United States are the top three countries. Respondents also came from Europe, Latin America, and Southeast Asia.

Top 20 countries

n = 1029

CountryCount%
India20519.9%
Nigeria14814.4%
United States of America838.1%
China363.5%
Brazil302.9%
France272.6%
Spain252.4%
Germany252.4%
United Kingdom252.4%
Canada181.7%
Argentina171.7%
Other171.7%
Vietnam171.7%
Russia171.7%
Poland151.5%
Kenya141.4%
Italy141.4%
Digital nomad131.3%
Turkey111.1%
Ukraine111.1%
Portugal101%
Bulgaria101%
Indonesia101%
Pakistan101%
Greece90.9%
Netherlands90.9%
Venezuela80.8%
Georgia80.8%
Armenia80.8%
Serbia70.7%
Ghana70.7%
Mexico70.7%
Hong Kong70.7%
Switzerland60.6%
Colombia60.6%
Bangladesh60.6%
Philippines60.6%
Sweden60.6%
United Arab Emirates60.6%
Thailand60.6%
Australia50.5%
Slovenia50.5%
Czechia50.5%
Japan50.5%
Taiwan50.5%
South Africa50.5%
Egypt50.5%
Morocco40.4%
Denmark30.3%
Uganda30.3%
Uruguay30.3%
Iran30.3%
South Korea30.3%
Cyprus30.3%
Lithuania30.3%
Hungary30.3%
Austria30.3%
Singapore30.3%
Romania30.3%
Ireland20.2%
Peru20.2%
Malaysia20.2%
Bolivia20.2%
Zambia20.2%
Israel20.2%
Sri Lanka20.2%
Honduras10.1%
Syria10.1%
Azerbaijan10.1%
Ivory Coast10.1%
Ethiopia10.1%
New Zealand10.1%
Chile10.1%
Bosnia and Herzegovina10.1%
Lebanon10.1%
Belgium10.1%
Belarus10.1%
Slovakia10.1%
Somalia10.1%
Ecuador10.1%
Uzbekistan10.1%
Cameroon10.1%
Zimbabwe10.1%
Costa Rica10.1%
Croatia10.1%
Togo10.1%
Latvia10.1%
Algeria10.1%
Tanzania10.1%

What is your native language?

n = 1030

English is the most common native language (34%), followed by Hindi, Spanish, Portuguese, and Russian.

Native LanguageCount%
English34633.6%
Hindi10610.3%
Spanish747.2%
Portuguese403.9%
Russian383.7%
French373.6%
German302.9%
Mandarin272.6%
Italian171.7%
Vietnamese171.7%
Yoruba161.6%
Turkish151.5%
Ukrainian151.5%
Chinese141.4%
Telugu131.3%
Urdu131.3%
Igbo111.1%
Tamil111.1%
Bengali101%
Arabic101%
Bulgarian90.9%
Swahili90.9%
Greek90.9%
Polish80.8%
Other80.8%
Indonesian80.8%
Armenian60.6%
Dutch60.6%
Serbian60.6%
Thai60.6%
Cantonese60.6%
Romanian50.5%
Malayalam50.5%
Hausa50.5%
Japanese50.5%
Swedish50.5%
Hungarian50.5%
Persian40.4%
Korean40.4%
Marathi40.4%
Georgian40.4%
Slovenian30.3%
Lithuanian30.3%
Filipino30.3%
Nepali30.3%
Danish20.2%
Punjabi20.2%
Kannada20.2%
Gujarati20.2%
Odia20.2%
Czech20.2%
Luganda10.1%
Amharic10.1%
Zulu10.1%
Sindhi10.1%
Kamba10.1%
Javanese10.1%
Konkani10.1%
Assamese10.1%
Tagalog10.1%
Cebuano10.1%
Slovak10.1%
Somali10.1%
Uzbek10.1%
Gikuyu10.1%
Hebrew10.1%
Shona10.1%
Setswana10.1%
Latvian10.1%
Berber10.1%

How old are you?

n = 1028

74% of respondents are between 18-34. The 25-34 bracket is the largest (38%), closely followed by 18-24 (36%). 9% of respondents are over 45.

How would you rate your coding ability?

n = 1033

57% read/write code professionally. 22% can read/write code but not professionally. 19% are still learning.

Which best describes your developer profile?

n = 786

The majority are smart contract developers (70%). Auditors/security experts are the second largest group (12%). Researchers, fullstack/backend developers, and tooling/library developers are also represented.

Which industry do you work in?

n = 990

Crypto is the most common industry (49%), followed by students (18%) and technology (16%).

How many years of professional coding experience?

n = 1006

Professional coding experience varies, with 3-5 years being the most common (23%). 25% have less than a year or no professional coding experience.

What is your primary programming language?

n = 978

Solidity is the most-used language (39%), followed by TypeScript (16%) and JavaScript (13%).

What is your favorite programming language?

n = 946

Solidity is the top favorite (32%), followed by Python (16%), JavaScript (11%), and Rust (10%).

What operating system do you use?

n = 976

Windows (38%), MacOS (31%), and Linux (29%) are the three main operating systems used.

vs. 2024: In 2024, MacOS led at 43%, followed by Windows (29%) and Linux (28%). In 2025, Windows leads at 38%, followed by MacOS (31%) and Linux (30%).

Solidity Usage

How often do you use Solidity?

n = 865

49% of respondents use Solidity daily, 32% weekly. 10% use it rarely or never.

How long have you been using Solidity?

n = 865

50% of respondents have 0-2 years of Solidity experience. 22% have 2-4 years. 4.5% have used Solidity for more than 8 years.

How would you rate your Solidity expertise?

n = 865

Self-rated expertise concentrates in the upper range: 7 and 8 are the most common ratings (34% combined). 14% rated themselves as beginners (1), forming a secondary peak.

Which Solidity versions do you use?

n = 871 | multiple choice

v0.8.x is the most widely used version (86%). Older versions account for 5-8% each (v0.4.x through v0.7.x).

Tooling & Infrastructure

How do you get the Solidity compiler?

n = 871 | multiple choice

79% get Solidity through a framework. npm (24%) and GitHub releases (16%) are the next most common.

What editor do you use?

n = 834

VSCode-based editors (VS Code, Cursor, Windsurf) account for ~80% of usage. Note: the survey grouped these under a single option, so we cannot distinguish between them. Other editors include Vim/Neovim, IntelliJ, and Zed.

What is your primary development framework?

n = 828

Foundry is the most used primary framework (57%). Hardhat v3 and v2 account for 33% combined.

vs. 2024: Foundry increased from 51% to 57%. Hardhat is at 33% combined in both years, but the 2025 survey distinguished between v2 (15%) and v3 (18%). Truffle dropped from 2.4% in 2024 to a single remaining user.

Which additional frameworks do you use?

n = 871 | multiple choice

Remix is the most used secondary framework (41% of respondents use it alongside their primary framework). 10% did not report using a secondary framework.

Which SDKs / libraries do you use?

n = 871 | multiple choice

ethers.js is the most used SDK (70%). viem follows (39%) and wagmi (33%). web3.py (16%) is also used. web3.js has 9 mentions. The library has been deprecated. alloy (6%) is also used, primarily in Rust-based tooling.

Compilation & Verification

Do you rely on older EVM version support?

n = 566

34% of respondents rely on compiler support older than the current EVM version (Osaka).

Oldest EVM version targeted

n = 162 | shown to those relying on older EVM support

Among those who rely on older EVM support, Paris, London, and Cancun are the most common targets.

Have you used the SMTChecker?

n = 854

45% of respondents don't know what the SMTChecker is. 30% have not used it. 18% have tried it and 7% use it often.

Have you used the IR pipeline (--via-ir)?

n = 852

35% of respondents don't know what the IR pipeline is. 28% use it sometimes, 15% use it often, and 22% do not use it.

vs. 2024: IR pipeline awareness also improved: 35% don't know what it is in 2025 (vs 46% in 2024).

Is the IR pipeline too slow to compile?

n = 284 | shown to IR pipeline users only

Among respondents who use the IR pipeline, 39% say it takes too long to compile.

Are you familiar with Sourcify?

n = 851

48% of respondents don't know about Sourcify. 28% know about it but don't use it. 23% know about it and use it.

vs. 2024: Sourcify awareness improved: 48% don't know about it in 2025 (vs 56% in 2024), and usage increased from 17% to 24%.

Contract verification pain points

n = 142 | multiple choice

Top themes: no issues (18%), framework verification tools (15%), verification flaky/unreliable (15%), metadata/compiler mismatch (12%), Etherscan issues (11%).

Etherscan is the de-facto standard and it's very unreliable and not transparent. Sourcify is much better and I wish I could tell Etherscan it's already verified there.
Too many chains, too many block explorers. Etherscan should use one source of truth i.e. Sourcify then verifying once should auto verify everywhere.

Do you use appendCBOR: false or bytecodeHash: none?

n = 589

49% don't know what it is. 11% use it.

Why do you use appendCBOR: false or bytecodeHash: none?

n = 39

Main reasons: easier verification (36%), deterministic/reproducible builds (28%), and reduced bytecode size (23%).

Chains & Deployment

Which chains do you deploy to?

n = 871 | multiple choice

Ethereum mainnet leads (52%). Base is second (40%), ahead of Arbitrum (35%) and Polygon (30%). 72 respondents (8%) mentioned deploying only to testnets.

Do you use other smart contract languages?

n = 871 | multiple choice

43% don't use any other smart contract language. Rust is the most selected alternative (28%). This may reflect cross-ecosystem usage (Solana, NEAR, Stylus) rather than EVM development, as the question did not restrict answers to EVM-compatible languages.

Developer Experience

How has the Solidity DevEx changed in the past year?

n = 521

73% report improvement (43% a bit, 31% a lot). 25% reported no change. 2% say it got worse (8 respondents).

vs. 2024: DevEx sentiment is slightly more positive: 73% report improvement (vs 67% in 2024). The percentage reporting things got worse is unchanged at 2%.

What has improved in the DevEx?

n = 151 | multiple choice

Top themes: personal skill growth (39%), Foundry improvements (15%), better tooling (14%), better compiler/errors (13%). Note: 39% of responses described personal skill growth rather than ecosystem improvements.

Foundry continued to ship solid features. Mainly improved tooling, not major language features that are super useful.
Custom errors in requires, custom storage layouts, transient storage.

Which recurring issues do you encounter?

n = 702 | multiple choice

Stack-too-deep errors are the most reported issue (47%). Bytecode size limits (33%) and debugging issues (33%) follow. Optimizer-related issues account for 13%. 23% report no recurring issues.

vs. 2024: The question format changed between years (single multi-select in 2024 vs separate checkboxes in 2025), which may account for some of the decrease. With that caveat: in 2024, stack-too-deep was reported by 68%, debugging by 55%, bytecode size by 51%, and optimizer issues by 22%. In 2025, these are 47%, 33%, 33%, and 13% respectively.

How could debugging be improved?

n = 79

Top themes: better error messages (28%), step-by-step debugger (16%), better stack traces (16%), transaction replay/simulation (14%), IDE integration (11%).

A step-by-step debugger similar to what you'd find in any modern IDE, with breakpoints and variable inspection for Solidity contracts.
Transaction replay with full state access. Being able to see every storage slot change and every call in sequence.

Which language features are most important?

n = 702 | multiple choice

Contracts as objects and inheritance is considered important by 54%, the most frequently selected feature. Unbounded dynamic arrays (34%) and inline assembly (29%) follow.

Biggest pain points with Solidity

n = 215 | multiple choice

Top themes from 215 free-text responses: stack-too-deep (20%), bytecode size limit (11%), gas optimization tradeoffs (9%), array/data structure limitations (8%), verbose syntax (8%).

Compiler's optimization defects continuously make me think about what's going on under the hood instead of focusing on my logic. Things that should be zero cost are not (functions, structs, variables).
Gas optimisations force bad programming practices, like repeating code inline instead of abstracting into functions, avoiding zeroing storage slots in favour of using arbitrary sentinel values.
Too many footguns around storage vs memory vs calldata, especially with complex structs, arrays, and inheritance.

How could the documentation be improved?

n = 102 | multiple choice

Most requested improvement is more real-world examples (27%). 16% say the docs are good as-is. Simplifying for beginners (14%) and better organization/search (12%) follow.

Better language reference page. Everything is explained in great detail, but it would be great with a reference page to quickly find what you need similar to: https://tour.gleam.run/everything/
I think the documentation should rationalize why Solidity is the way it is more often - and how it relates to the EVM's design.

Most wanted near-term features

n = 702 | multiple choice

Better gas optimizations is the most requested feature (44%). EIP-712 typehash support (29%) is the second most requested. Reference types in transient storage (23%) and more fine-grained storage layout control (23%) are also among the top requests.

Core Solidity

204 respondents are familiar with Core Solidity (30% of the 678 who answered this question).

Are you familiar with Core Solidity?

n = 678

Core Solidity is an initiative to redesign the language from scratch, incorporating lessons from 10 years of Solidity development. 30% of respondents who answered this question are familiar with it.

Which Core Solidity features interest you most?

n = 204 | multiple choice | shown to those familiar with Core Solidity

Among those familiar, better error handling / try-catch replacement (43%) and better delegatecall mechanism / library replacement (41%) are the most selected. 14% selected none.

Would removing inheritance cause challenges?

n = 200 | shown to those familiar with Core Solidity

63% say removing inheritance would cause challenges.

How difficult would rewriting to traits be?

n = 201 | shown to those familiar with Core Solidity

44% estimate the rewrite to type classes/traits would be somewhat difficult, 21% very difficult.

Would your codebase benefit from compile-time evaluation?

n = 201 | shown to those familiar with Core Solidity

71% of those familiar with Core Solidity say their codebase would benefit from improved compile-time evaluation.

Feedback on Core Solidity

n = 43 | multiple choice | shown to those familiar with Core Solidity

23% are skeptical/cautious. 21% are excited/supportive. 21% are concerned about fragmentation. 12% want it kept low-level. "Security focus" refers to respondents who emphasized that security should be the primary goal of any language redesign.

Excited to see this project unfold. If we can have a new language that takes all the pain points of the last 10 years of Solidity and fixes them it would be great.
I am very concerned that Core Solidity is trying to turn solidity into some kind of academic Haskell language that requires a PhD in types to use.
Still unclear how core solidity is not a new language and how the syntax and semantics will be made backwards compatible.
I care most that the bytecode is good. Syntactic sugar is less important to me.
Please do not remove inheritance and contracts as classes, everything should be backwards compatible as we will lose all the years of work, libraries and standards created.

AI in Solidity Development

Do you use AI tools for development?

n = 650

58% use AI tools daily. 88% use them at least monthly. 4% don't use AI and don't plan to.

How do you view AI in development?

n = 650

77% of respondents view AI favorably (46% favorable, 31% very favorable). 9% hold unfavorable views. 14% are indifferent.

How much do you trust AI-generated code?

n = 591

49% somewhat trust, 30% somewhat distrust, 15% highly distrust, and 6% highly trust AI output. Notably, adoption outpaces trust: 88% use AI tools at least monthly, yet 45% express some level of distrust in the output.

What do you use AI for in your workflow?

n = 629 | multiple choice

Testing (61%), documenting code (59%), and learning about codebases (58%) are the top AI use cases. Debugging (53%), auditing (52%), and searching for answers (52%) follow. Writing code (49%) and reviewing code (49%) are somewhat lower.

Which editors do you use with AI agents?

n = 629 | multiple choice

VSCode (60%), Cursor (32%), and Antigravity (16%) are the top editors used with AI agents.

Preferred AI assistant

n = 545

Claude Code is the most selected preferred assistant (61%), followed by Codex/ChatGPT (16%) and Gemini (10%). Note: this survey was distributed via Solidity community channels, which may introduce distribution bias.

Cross-Analysis

For the cross-analysis charts, self-rated expertise was grouped into three tiers: Beginner (1-4), Intermediate (5-7), and Expert (8-10).

Self-rated expertise by years using Solidity

n = 865

Self-rated expertise increases with years of Solidity experience. The most common self-rating of 8+ appears at the 4-6 year mark.

Self-rated expertise by years coding

n = 865

Self-rated Solidity expertise also correlates with general programming experience, though less strongly than with Solidity-specific experience.

Recurring issues by expertise level

n = Beginner (1-4): 181, Intermediate (5-7): 285, Expert (8-10): 236

Stack-too-deep is reported by 25% of beginners vs 65% of experts. Bytecode size limit follows a similar pattern (17% vs 47%). Debugging issues are consistent across all levels (29-35%).

Expertise distribution by framework

n = Foundry: 467, Hardhat v3: 145, Hardhat v2: 126, Remix: 62

Framework choice correlates with expertise level: Remix users are predominantly beginners (60%), Hardhat v3 has the highest share of intermediate users (47%), and Foundry has the highest share of expert users (38%).

AI usage by expertise level

n = Beginner (1-4): 164, Intermediate (5-7): 260, Expert (8-10): 226

Daily AI usage is higher among more experienced respondents: 52% of beginners, 58% of intermediate, and 62% of expert respondents use AI daily.

AI trust vs. usage frequency

n = 591

36% of daily AI users express distrust (somewhat or highly) in the output. Nearly all who highly trust AI (37 of 38) are daily users.

Expertise: students vs. professionals

n = Students: 147, Professionals: 692

Students cluster at lower expertise (peak at 5), professionals peak at 7-8.

Framework choice: students vs. professionals

n = Students: 137, Professionals: 666

Remix usage is higher among students; Foundry is more common among professionals.

AI usage: students vs. professionals

n = Students: 101, Professionals: 536

AI adoption is similar between students and professionals.

Final Feedback Highlights

The 151 final feedback responses included the following recurring themes:

Feature requests:

  • Bytecode size limit increase (mentioned multiple times)
  • Generics support for library developers
  • Better type conversions
  • Native cryptographic primitives and smoother Yul integration
  • Pre-dispatch hook: ability to run code before/after method dispatch
  • Development tools for Zed editor

AI-related:

  • Multiple respondents report AI-generated Solidity is unreliable
  • Request for the Solidity team to help AI write more secure code

Community and communication:

  • More visibility and outreach for Solidity
  • More detail in Core Solidity article on try-catch replacement and typeclasses
  • More outreach for the survey through ecosystem projects
Solidity is a surprisingly good language for expressing the kinds of problems you have for EVM smart contracts. Thanks for it.
Smart contracts are hard since you have to balance the quality of the code vs the cost for the user. Ideally correctly written code should also be the proper way to optimize for gas, and not the other way around.
Please fix the bytecode size limit, it's not always practical to refactor contract into multiple smaller ones, doing so is always a huge lift in dev and ops work.
Solidity and the surrounding ecosystem are moving in the right direction, especially in tooling, testing frameworks, and developer workflows. However, developer experience is still heavily constrained by debugging limitations, EVM-level abstractions, and low-level complexity that slows productivity and increases risk.
Keep pushing for better native cryptographic primitives and smoother Yul integration. It makes building privacy-focused tools and DeFi mechanisms safer and gas-efficient.
It feels like the primarily direction of Solidity should be security, compiler performance, more aggressive bytecode optimization + developer ergonomics.
Keep pushing, everyone. It's a pleasure to write smart contracts today compared to where we were 3 to 5 years ago.

Methodology

  • Survey ran February-March 2026 using LimeSurvey
  • 1,482 total submissions; 387 responses dropped (empty submissions or those who did not complete any survey page)
  • 640 fully completed (all 8 pages), 1,095 usable (reached page 1+)
  • Free-text "Other" answers were normalized and merged into parent columns
  • Multi-value free-text answers are counted toward each mentioned item
  • Troll/noise entries were filtered out
  • Per-question N is used throughout: each chart shows the number of respondents who answered that specific question
  • Questions marked "conditional" were only shown to respondents who met a prerequisite
  • Distributed via Solidity community channels and social media
  • Free-text responses were manually categorized by theme. A single response can match multiple themes. Misinterpretation or miscategorization of individual responses is possible; treat theme counts as approximate

Get involved

GitHub

Twitter

Mastodon

Matrix

Discover more

BlogDocumentationUse casesContributeAboutForum

2026 Solidity Team

Security Policy

Code of Conduct