Coinbase draws the line: “Artificial intelligence is not a choice, but an obligation”
When Coinbase CEO Brian Armstrong declared he was firing engineers for refusing even to engage with AI tools — it was perceived as a shockwave. After purchasing corporate licenses for GitHub Copilot and Cursor, he took a step rarely seen in tech companies: he set a strict deadline — one week for all engineers to complete onboarding, otherwise — a Saturday meeting with explanations. Armstrong admitted the approach was “harsh,” but deemed it necessary to make clear — deviations from strategy were unacceptable.
At that meeting, it became clear that some employees were genuinely away from the office — on vacation or blocked by objective circumstances — and their departure was arranged softly. Others, however, did not provide valid reasons — for them, work at Coinbase ended.
The fact in numbers: already around 33% of Coinbase’s code is created with AI, and the company aims to raise this figure to 50% by the end of the quarter. In parallel, regular “AI Speed Run” sessions have been launched, where engineers share experiences applying AI across different stages of development.
This case is not just an internal story but a vivid signal: major companies increasingly see artificial intelligence not as a tool but as standardized infrastructure. Coinbase chose a direct and intensified acceleration of this transition, even at the cost of staff losses. Consensus or comfort yield to speed and determination.
