Coinbase CEO: AI Makes Engineers Super Builders
Coinbase CEO Brian Armstrong stated that AI tools like Cursor and custom agents have generated over 40% of the company's code, with engineers increasing their merged PR count by 55% and saving 7 hours of manual work per week.
The company restructured and laid off employees in May to form an AI-native team, resulting in a 78% year-on-year increase in PRs per engineer, with a goal of reaching 50% AI-generated code by October.
Armstrong views this as a path for the AI and crypto integration to surpass competitors, but critics question whether speed compromises code quality, while issues at the exchange and stock performance lag behind.
ABAB AI Insight
Coinbase is restructuring its engineering processes through AI, significantly boosting productivity. Armstrong's statement continues his positive stance on AI adoption, aiming to transform the company into an AI-driven crypto platform.
On the capital front, AI tools reduce development costs and accelerate iteration. Strategically, Coinbase is optimizing its workforce structure through layoffs and reorganization, aiming to maintain a lead in the competitive crypto services market.
Similar to other tech companies adopting AI coding tools, the Coinbase case illustrates the disruptive impact of AI on software engineering, with productivity surges accompanied by debates over quality and employment structure adjustments.
Essentially, this represents technological substitution, where AI agents compress engineering labor demand, concentrating capital on AI-native teams and efficient development processes, driving rapid upgrades in crypto industry infrastructure.
ABAB News · Cognitive Law
The higher the proportion of AI-generated code, the more engineers transition from coders to architects.
When productivity surges, quality control becomes the new bottleneck.
The speed of AI transformation in tech companies determines the competitive landscape of the industry.