DeepSeek Tops Ramp's U.S. Corporate Spending Rankings as American Firms Embrace First Wave of 'Data-Direct-to-China' Procurement
0xBroomberg
Chinese AI startup DeepSeek topped Ramp's trending-software-vendor chart in June, with some US companies paying DeepSeek directly and routing data through Chinese-hosted servers — the strongest signal yet that firms are hunting for cheaper alternatives to OpenAI.
What does this ranking actually measure?
Ramp tracks US corporate software spending. Its "trending vendor" chart counts the number of companies making a first-ever payment to a given vendor.
This means → DeepSeek's No. 1 spot shows it attracted the most new corporate buyers in June — not that its total revenue surpassed rivals.
In plain terms = the chart measures "who is the biggest new-customer magnet," not "who earns the most."
Why does paying DeepSeek directly matter?
Ramp chief economist Ara Kharazian noted that direct payments suggest these firms are not self-hosting DeepSeek's open-source models on their own servers — they are sending and receiving data through DeepSeek's API.
This means → US corporate data is traveling back and forth through servers hosted in China, and the companies are knowingly accepting the compliance and security risk.
This reflects a trade-off: with AI service costs running high, some firms have decided the savings are worth the data-sovereignty risk.
How large is DeepSeek's actual share?
The absolute share remains tiny. DeepSeek briefly hit 0.3% adoption on Ramp's AI index in January 2025, fell back to 0.1%, and stayed at 0.1% through April.
Anthropic and OpenAI, by contrast, hold 34.4% and 32.3% respectively — together commanding more than two-thirds of the index.
In plain terms = DeepSeek's footprint in the US enterprise market is still negligible, but the direction matters more than the absolute number — going from near-zero to the top of the new-customer chart happened fast.
What is the real signal here?
Kharazian's own words: "This is perhaps the biggest signal that companies are looking for cheaper alternatives to OpenAI and Anthropic."
This means → the key variable is not how strong DeepSeek is — it is that US companies' tolerance for high AI costs is hitting a ceiling. When price pressure is high enough, even the data-sovereignty red line becomes negotiable.
This reflects a deeper tension in the AI industry: the pricing power of leading models is cracking against enterprises' will to cut costs — the gap is now visible.
Content is for reference only, not financial advice.