Coinbase has intensified its reliance on artificial intelligence in software development, with nearly 40% of its daily code now generated by AI systems.
Chief Executive Brian Armstrong stated that the cryptocurrency exchange aims to push this figure beyond 50% by October, demonstrating the company’s commitment to integrating AI at every layer of its engineering process.
Armstrong’s Hardline on AI Adoption
Sponsored
In a recent post on X, Armstrong emphasized that AI adoption is a strategic imperative for Coinbase. This ambition follows months of internal initiatives aimed at standardizing coding assistant usage, implementing new developer tools, and evaluating productivity through both quantitative and qualitative metrics.
“~40% of daily code written at Coinbase is AI-generated. I want to get it to >50% by October,” Armstrong wrote.
Armstrong’s focus on AI extends beyond voluntary exploration; it has become a mandate. He reported that engineers were required to onboard AI tools like GitHub Copilot and Cursor. Those who resisted were challenged to give reasons for their stance, and in some instances, this led to dismissals.
“AI’s important. We need you to all learn it and at least onboard,” he stated, reflecting a decisive strategy to expedite the adoption process.
Coinbase has also introduced monthly “AI speedruns,” where employees showcase the practical application of AI in their workflows. Armstrong revealed that about one-third of the company’s code currently comes from AI, with aspirations to achieve 50% by year’s end.
Developer Productivity Framework
Sponsored
On August 6, Coinbase further explained its strategy in a blog post by Kyle Cesmat and Chitra Venkatramani. They outlined a framework based on DORA metrics and developer surveys, emphasizing the responsible integration of AI. The blog acknowledged that certain systems are unsuitable for AI-generated code due to security sensitivities, highlighting the need for caution.
More than 1,500 engineers reportedly use AI-powered tools daily, benefiting from IDE integrations and custom model routers. Cursor has become a favored integrated development environment, although alternatives like Cody and JetBrains remain part of the developer toolkit.
Coinbase actively tracks AI adoption through metrics such as lead time to change, deployment frequency, and the proportion of AI-generated lines of code. The firm anticipates that the output from AI will soon surpass that generated by human developers.
Big Tech Leaders Highlight Similar Shifts
Sponsored
Coinbase’s proactive stance on AI mirrors a wider trend observed among leading technology companies. For instance, Google’s Sundar Pichai has indicated that over 25% of new code is produced by AI, calling this a fundamental shift in software development.
Meanwhile, Microsoft’s CEO, Satya Nadella, reported that up to 30% of his company’s code is now AI-generated, predicting that this figure could reach 50% in the next couple of years.
Such statistics illustrate that AI has evolved from being a mere experimental tool to an essential element driving productivity within corporate technology strategies.
Research Confirms Global Adoption
Academic studies bolster these corporate claims. A 2025 study on arXiv analyzed millions of GitHub repositories, finding that AI-generated code constituted approximately 30.1% of functions in the United States, with figures of 24.3% in Germany and 23.2% in France.
Moreover, the research estimated that productivity gains from AI-generated code could create an additional economic value of $10 billion to $14 billion annually in the U.S. alone.
This analysis draws attention to a significant caveat: the adoption of AI-driven development is unevenly distributed. More advanced economies and tech hubs exhibit higher levels of adoption, while other regions lag behind.
Developers Report Daily Use
Sponsored
Surveys highlight the growing embrace at the developer level. The Stack Overflow 2025 Developer Survey revealed that 84% of respondents either use AI coding tools or plan to adopt them, with over half reporting daily usage. Notably, more than 40% of these developers indicated that at least half of their codebase is now AI-generated.
Coinbase’s goal of exceeding 50% AI-generated code by October exemplifies a larger shift within the industry. Companies from Silicon Valley to Europe are increasingly embedding AI into their development processes, fundamentally changing the interaction between human and machine contributions.
As the demand for productivity rises, so too does the challenge to enhance outcomes without sacrificing quality or security. The company’s blog states, “We are building a culture of innovation to make Coinbase the best place to build.”