Coinbase's Head of Platform, Rob Witoff, said 95% to 100% of the company's code is now written with the help of AI, per an exclusive interview. Most engineers at the exchange are running five to ten AI agents concurrently, producing work the platform lead said equates to roughly 1,200 employees of output.
Why it matters
Coinbase is a US-listed, S&P 500-adjacent financial infrastructure company, not a software developer pitching a productivity story. A senior platform lead putting a near-100% number on the record is the cleanest senior-executive signal yet that AI-coded software has crossed from experiment to default at a regulated venue. The implication is not just faster shipping but a fundamentally different cost-per-feature curve for any company running engineering at scale.
Market impact
The productivity baseline in software has now been re-rated by an institution investors can underwrite. Exchanges, banks and large fintech employers all run comparable engineering organisations, and any of them credibly claiming a similar AI-share figure would compress the human-engineer multiplier that has supported headcount-rich valuation models in the sector. Watch for the next major listed venue to publish an analogous number, because the comparison will drive the multiple.
Frequently asked questions
-
What did Coinbase's Head of Platform say about AI coding?
Rob Witoff said 95% to 100% of Coinbase's code is now written with the help of AI, with most engineers running five to ten AI agents at a time.
-
How many employees' worth of output does the AI coding effort produce?
Witoff said the AI agents collectively produce coding work equivalent to roughly 1,200 employees.
-
Why is this announcement investable signal rather than a tech demo?
Coinbase is a US-listed, regulated financial infrastructure company, so a near-100% AI-code figure from a senior platform lead re-rates the productivity baseline other listed exchanges and fintechs will be measured against.
-
Does this affect publicly traded peers and competitors?
Any other listed venue or large fintech that publishes an analogous AI-share number will be benchmarked against Coinbase's 95–100% figure, which can compress the human-engineer multiplier in valuation models.
-
What should investors watch next?
The next major listed exchange or fintech employer to disclose an equivalent AI-coding share figure is the catalyst, because the comparison against Coinbase will likely move the sector's multiple.
CoinTelegraph