The most popular AI tool among US businesses this month did not come from San Francisco. It came from Hangzhou.
DeepSeek, the Chinese AI startup, is making a real impact and gaining commercial momentum. According to Ramp, a New York-based corporate spending platform tracking payments from more than 50,000 US businesses, DeepSeek topped its June trending vendor index, looking back to May, which measures when companies pay a software vendor for the first time.
The Cost Problem Driving the Switch
It appears Deepseek’s revenue model is shining compared with US rivals. Anthropic’s incentives are structurally misaligned with cost-conscious businesses.
The company makes more money when businesses purchase more tokens, pushing users toward expensive models even when cheaper options would suffice. Uber’s CTO announced the company had already blown through its entire 2026 AI budget.
DeepSeek recently cut its V4 Pro model price by 75%, after which benchmark firm Artificial Analysis ranked it among the world’s best on an intelligence-per-dollar basis. On legal AI benchmarks, it ranked just below GPT-5.5 and was deemed clearly viable for professional workloads.
US Data at Risk from DeepSeek
Crucially, rather than self-hosting DeepSeek’s open-source models, US firms are paying the company directly and sending real business data through its servers in China.
“In probably the biggest sign that companies are looking for cheaper alternatives to OpenAI and Anthropic, some are willing to use cheaper, Chinese models, sending US data back and forth from China-hosted servers,” said Ara Kharazian, lead economist at Ramp Economics Lab.
The IPO Problem Makes It Worse
Anthropic filed for an IPO valued at approximately $965 billion on June 1. OpenAI closed a $122 billion funding round in March at an $852 billion valuation. At those numbers, neither company can realistically compete on price with a startup that just slashed its rates by 75%.
For now, DeepSeek’s overall market share remains a fraction of its American rivals. But when enterprise AI budgets run dry, and a cheaper alternative clears professional benchmarks, the flag on the server stops mattering as much as the bill at the end of the month.