business
Coinbase CEO Outlines Five Steps to Cut AI Spending While Raising Usage
Coinbase CEO Brian Armstrong detailed five tactics to lower AI costs on Friday. The steps include defaulting to cheaper open-weight models from Chinese labs and giving engineers visibility into their token spend. Business Insider reported the post came less than two months after the company cut 14 percent of staff.
Audio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Coinbase CEO Brian Armstrong posted five strategies on X on Friday to reduce the company's AI spending while allowing engineers to increase token use. Business Insider reported that token usage at the firm recently hit one of its highest levels, yet AI spending has dropped to nearly half its peak.
ai and Kimi 2.7 from Moonshot AI through its LLM gateway.
He added that engineers are still encouraged to select the model best suited to each task. The company routes prompts to different models according to difficulty, directing frontier models to planning work and lighter models to execution. Armstrong wrote that humans should eventually stop choosing models and let AI handle the selection.
Additional measures include improved caching to lower inference costs, instructions to start new sessions when switching tasks to keep context lean, and full visibility for every engineer into individual token usage and spend. Armstrong stated that Coinbase expects greater output from employees who spend more on AI. "The goal isn't to suppress usage.
It's to build the infrastructure that makes exponential growth sustainable," he wrote. Business Insider noted the post followed a May 2026 statement in which Armstrong said engineers using AI now complete in days what once required a team weeks. Coinbase reduced its workforce by 14 percent less than two months before the Friday post.
