A coalition of more than 200 startups is pushing back against proposed U.S. restrictions on Chinese AI models, arguing the move would stifle innovation and raise costs across the tech sector — including in crypto. The opposition comes as Coinbase disclosed this week that it slashed its AI costs by 50% after switching to the Chinese model Kimi, underscoring the financial stakes for companies that rely on affordable AI tools.
Coinbase's Kimi bet
Coinbase, one of the largest U.S. crypto exchanges, said it cut AI-related expenses in half by using Kimi, a model developed by a Chinese startup. The savings are significant for a company that has been under pressure to trim costs amid a prolonged crypto winter. Coinbase's move also highlights how deeply Chinese AI has already been integrated into American crypto infrastructure — a fact that could complicate any regulatory crackdown.
Why startups are pushing back
The debate in Silicon Valley centers on whether to restrict Chinese AI startups from operating in the U.S., or to limit American companies' access to Chinese AI models. More than 200 startups have signed a letter opposing such restrictions, arguing they would hurt competition and raise barriers for smaller players. The crypto industry, which often relies on lean operations and open-source tools, would be particularly exposed if cheap Chinese AI models were cut off.
What's at stake for crypto
Proposed restrictions on Chinese AI could ripple through the cryptocurrency industry, where AI is used for everything from trading bots to fraud detection to customer service. If American companies lose access to models like Kimi, they may face higher costs or slower development — at a time when the sector is already fighting for margin. The debate is still in its early stages, but the coalition of startups is making clear that any ban would face fierce resistance.
The next concrete step is unclear. The White House has not yet proposed formal rules, but the issue is expected to come up in upcoming trade and tech policy discussions. For now, the 200-plus startups are waiting — and watching.
