Coinbase CEO Brian Armstrong wants artificial intelligence to write at least half of the company’s code by October, marking a bold shift in how the crypto exchange builds its platform.
Key Takeaways
- Over 40% of Coinbase’s code is now generated by AI, with a goal of reaching 50% by October 2025.
- CEO Brian Armstrong mandated all engineers to onboard AI tools like GitHub Copilot and Cursor, firing those who refused.
- Coinbase is hiring aggressively for AI-skilled developers, with nearly half of its open roles in engineering.
- Armstrong believes AI is not replacing jobs but accelerating productivity, allowing single engineers to complete work in days instead of months.
What Happened?
Coinbase has embraced artificial intelligence as a cornerstone of its development process, with over 40% of the code now being written by AI tools. CEO Brian Armstrong wants that figure to hit 50% within weeks and has taken decisive steps to ensure compliance across the engineering team.
~40% of daily code written at Coinbase is AI-generated. I want to get it to >50% by October.
— Brian Armstrong (@brian_armstrong) September 3, 2025
Obviously it needs to be reviewed and understood, and not all areas of the business can use AI-generated code. But we should be using it responsibly as much as we possibly can. pic.twitter.com/Nmnsdxgosp
AI-Powered Coding: The New Coinbase Norm
Brian Armstrong is not just dabbling with AI. He is fully committing Coinbase to a future where machine-generated code is the norm. Since April, the percentage of AI-written code at the company has doubled. Armstrong credits tools like GitHub Copilot, Claude Code, and Cursor for enabling this transformation.
These tools allow developers to:
- Refactor existing codebases faster
- Create new applications in days
- Enhance code quality with fewer bugs and redundancies
According to Armstrong, “This has enabled profound success stories that weren’t possible 12 months ago.”
Mandatory AI Onboarding for Engineers
The AI push is not optional at Coinbase. Armstrong has issued a clear directive: all engineers must onboard AI coding tools immediately. In an internal Slack message, he told staff to start using the tools by the end of the week. Those who failed to comply were asked to explain themselves in a weekend meeting.
Fortune reported that during the Saturday review, some engineers were fired for failing to adopt AI tools without a valid reason. The company later confirmed it had purchased enterprise licenses for Copilot and Cursor to scale up AI usage.
Armstrong’s stance is echoed by tech leaders like Satya Nadella of Microsoft, who recently said 20% to 30% of Microsoft’s code is AI-generated.
Hiring Focused on AI Talent
Despite public fears that AI may lead to job cuts, Coinbase continues to expand its workforce. Out of approximately 350 current job openings, nearly 150 are for technical and engineering roles, many of which reference AI capabilities directly in the descriptions.
- 93 roles are back-end engineering positions
- 56 roles are in customer experience
- AI-specific roles are on the rise across departments
While broader crypto industry hiring has cooled, Armstrong is doubling down on acquiring talent that can work alongside AI.
Broader Industry Trend
Armstrong is not alone in betting on AI. Leaders at Google, Shopify, and Galaxy Digital are pushing similar strategies. Galaxy CEO Michael Novogratz believes AI agents will become major users of stablecoins, predicting explosive growth at the intersection of crypto and artificial intelligence.
In parallel, a rise in “vibe coding” has taken off among new developers. With tools like Copilot, even non-coders are creating apps by simply prompting AI, bypassing traditional programming education.
CoinLaw’s Takeaway
In my experience, few tech leaders have gone as all-in on AI as Brian Armstrong. It’s rare to see a CEO not only embrace automation but actively fire those who resist it. That tells me Coinbase isn’t just experimenting with AI, it’s restructuring itself around it. I found it especially telling that single engineers are now doing in days what used to take teams months. That’s not just efficiency, that’s transformation. If you’re in crypto or tech and not learning AI, this is your wake-up call.
