Coinbase CEO cuts AI spending in half as crypto activity shifts, frames Bitcoin pullback as a “cool breeze”

Coinbase Chief Executive Brian Armstrong said the company has halved its spending on artificial intelligence, arguing that the move better matches current demand as token usage rises. In remarks reported by Yahoo Finance, Armstrong linked the cost adjustment to how Coinbase is using AI tools, pointing to changing usage patterns rather than a broad retreat from the technology.

The comments arrive during a period when the crypto market has been under pressure. Armstrong described Bitcoin’s bear-market phase as a “cool breeze,” language that suggests he does not view the downturn as fundamentally derailing the business. That stance, if reflected in day-to-day strategy, implies Coinbase is trying to manage costs while keeping optionality for future trading and product volumes.

Armstrong’s spending adjustment centers on AI costs. According to the report, Coinbase cut AI-related expenditures by about half, while token usage increased. “Token usage” in this context refers to the volume of text or data processed through AI systems, which can scale with customer support demand, internal workflows, or other AI-assisted services. A rise in token usage can raise AI operating expenses, which makes cost controls particularly salient.

The CEO’s framing also indicates that Coinbase is treating AI as a controllable operating line item rather than a fixed bet. By reducing AI spend even as usage rises, the company is likely trying to optimize unit economics, such as improving how AI models are deployed, adjusting infrastructure consumption, or refining processes to reduce waste. The report does not provide the specific mechanisms Coinbase used, only the direction of spending and the relationship to token usage.

Coinbase’s position in the crypto sector adds urgency to such cost management. Trading volumes and crypto asset prices can change quickly, affecting fee revenue and the overall pace of retail and institutional participation. In that environment, companies that rely on variable market activity often respond by tightening expenses and emphasizing efficiency, especially in support operations and technology development.

While Armstrong spoke about AI costs and Bitcoin’s downturn, the report did not lay out additional financial detail such as the absolute dollar amount of the AI reduction, the timeframe for the cut, or how the company expects AI spending to evolve over the next quarter or year. It also did not specify whether the AI budget includes research and engineering, customer-facing features, or third-party model usage. Those gaps matter because investors and analysts typically want to understand whether the cut is temporary, tied to experimentation, or part of a longer-term restructuring of technology spending.

Still, the overall message is clear: Coinbase is aligning technology spending with current operating realities rather than allowing AI costs to expand unchecked. If token usage is increasing while spending is falling, that implies at minimum that the company has found ways to deliver similar or improved output at lower cost, or that some AI services are being scaled more efficiently. The report also suggests Armstrong believes the market’s decline will not prevent Coinbase from executing its plans.