Coinbase has cut its AI spending in half even as employee usage of the tools keeps rising, CEO Brian Armstrong said, according to The Information. It’s a counterintuitive headline for an industry where the assumption has been simple: more AI use means bigger bills. Coinbase is reporting the opposite.
What stands out here is the direction of the trend. Usage up, cost down. That combination tells you something about where the economics of AI are heading.
What Armstrong Said
The crypto exchange’s chief executive framed the shift as a sign that AI tooling is getting cheaper to run at scale, even as it becomes more deeply embedded in how the company works. The Information reports that Coinbase managed to roughly halve its AI costs while adoption inside the company grew.
Armstrong has been vocal about pushing AI across Coinbase. Earlier this year he made headlines for telling engineers to adopt AI coding tools or risk being let go. So this update lands as a progress report on that bet: the company leaned in hard, and now it’s saying the spending is coming down, not up.
Why the Cost Is Falling
The headline number reflects a broader force reshaping the AI market. A few things are happening at once:
- Model prices keep dropping. The cost per token from major providers has fallen sharply over the past year. The same task that was expensive in early 2025 is a fraction of the price now.
- Smaller models are catching up. Companies are learning they don’t need the biggest, most expensive model for every job. Routing simple tasks to cheaper models cuts the bill fast.
- Better engineering. Caching, smarter prompting, and tighter integration mean fewer wasted calls. Teams that have lived with these tools for a year know how to use them efficiently.
Put those together and you get exactly what Coinbase is describing. The company isn’t using AI less. It’s using it smarter, and the market is handing it lower prices at the same time.
Why It Matters
This is significant because it pushes back on one of the loudest fears in enterprise AI: that costs will spiral as adoption spreads. Finance teams have been bracing for AI to become a runaway line item. Coinbase is offering a real example where the curve bent the other way.
For practitioners and executives, the takeaway is practical. If your AI bill is climbing as fast as your usage, that may be an engineering and procurement problem, not an inevitability. The tools to control spend exist: model routing, caching, usage monitoring, and renegotiated provider rates.
There’s also a competitive signal here. Coinbase is positioning AI efficiency as an operational advantage, not just a feature. A company that can scale AI usage while shrinking its costs gets to do more with the same budget. That’s leverage.
The Bigger Picture
The status quo for most of the past two years was straightforward. AI was treated as a strategic investment where rising costs were the price of staying current. Many firms accepted growing bills as proof they were serious.
Coinbase is challenging that framing. The message is that mature AI adoption can actually lower unit costs over time, the same way cloud computing eventually did once teams learned to manage it.
Expect more companies to start reporting numbers like these as the market matures. The early phase of AI spending was about access and experimentation, and budgets reflected that. The next phase is about efficiency, and the leaders will be the ones who can show usage and cost moving in opposite directions.
Whether Coinbase’s halving holds as usage keeps growing is the question worth watching. If it does, it becomes a template other CFOs will want to copy. You can find the full details in the original report from The Information.