Coinbase has essentially replaced most of its engineering workforce with AI, at least on paper. The crypto exchange now reports that between 95% and 100% of its code is either written by or assisted by large language models, a figure that sat at just 40% as recently as February 2026.

From 5.7% to nearly everything

The progression tells the real story. In the first quarter of 2025, AI-generated code accounted for just 5.7% of Coinbase’s total codebase. By the fourth quarter of 2025, that share had crossed 50%. By the first quarter of 2026, it had reached roughly 80%. Now, the company says it is approaching full AI assistance across the board.

Every employee at Coinbase reportedly uses AI tools on a daily basis. The efficiency gain the company is pointing to is significant. Coinbase says its current AI deployment is equivalent to the output of approximately 1,200 human developers. The longer-term ambition is more striking: the company believes it could scale that capacity to match the output of 100,000 human developers by 2030.

Layoffs followed the math

In May 2026, Coinbase announced a roughly 14% reduction in its workforce, cutting approximately 700 employees. The company framed the decision as a need to operate with startup-level agility, using AI as a foundational part of its operations rather than a supplementary tool.

CEO Brian Armstrong has been tracking AI integration targets internally and pushing the organization toward progressively higher benchmarks since 2025. The restructuring also appears to have changed what the remaining engineering roles look like. Senior engineers are now managing multiple AI agents rather than writing code line by line, shifting the human role from producer to supervisor. Internal prototyping has moved to near-complete automation.

Core cryptography still requires substantial human review, reflecting a reasonable distinction: you can let an AI generate a new dashboard feature with relatively limited downside risk, but you probably want humans closely watching anything that touches private key management or cryptographic protocols securing user funds.

What this means for Coinbase investors and the broader industry

From a cost-structure perspective, a company that can do the work of 1,200 developers without hiring 1,200 developers is operating with a fundamentally different cost base than its competitors. From a risk perspective, the speed of this transition raises questions. Moving from 40% to near-100% AI-assisted code in a matter of months is fast, even by tech industry standards. Mandatory human review of all AI-produced code is the safeguard Coinbase is pointing to, and it matters that they are maintaining that requirement, especially in a regulated financial services context where a software bug is not just a bad user experience but a potential compliance event or security breach.