From TokenPost · · 1m 10s
AI Coding Data Shows Widespread Use but No Industrywide Code Share
AI tools now generate 42% of code for developers, but no single industrywide figure exists for total output.
In brief
AI tools now generate 42% of code for developers, but no single industrywide figure exists for total output. Professional developers estimate AI generated 42% of their code in 2025, rising to 65% by 2027, but no industrywide code share exists. Originally reported by TokenPost.
AI Code Use Estimates
Professional developers estimate AI generated 42% of their code in 2025, rising to 65% by 2027.
“42% for 2025, 55% for 2026, 65% for 2027”
Google's Internal Figure
At Google, AI-generated code approved by engineers accounted for 75% of new code on April 22, up from 50% the previous fall.
“75% of new code on April 22”
Organizational Output Measure
67% of 200 U.S. technology decision-makers placed AI-generated or refactored code at 51% to 75% of their weekly output.
“51% to 75% of weekly output”
Global Adoption Survey
90% of software developers used AI tools at work, with a median of two hours of daily use.
“90% used AI tools, two hours daily”
Accuracy and Trust
46% of developers distrust AI output accuracy, while 33% express trust in it.
“46% distrust, 33% trust”
Code Change Patterns
Duplicated code blocks per million changed lines rose 81% from 40.3 in 2023 to 73.0 in 2026 year to date.
“81% rise from 40.3 to 73.0”
At a glance
Professional developers estimate AI generated 42% of their code in 2025, rising to 65% by 2027, but no industrywide code share exists.
