US firms – Some US companies are starting to pay directly for China’s DeepSeek as enterprise AI budgets tighten, with DeepSeek topping Ramp’s June list of trending software vendors. The movement is still small next to OpenAI and Anthropic, but it signals growing willingn
For some US companies, the appeal of Silicon Valley AI tools is starting to feel expensive in the real world—so they are testing something else.
In June. DeepSeek climbed to the top of Ramp’s list of “trending software vendors. ” a ranking tied to when businesses first purchase from a software vendor. DeepSeek’s appearance on that list points to new spending on the Chinese AI startup. even if it remains far from reshaping the market dominated by OpenAI and Anthropic.
Ramp is a New York-based corporate spending platform, and its tracking matters because it moves beyond headlines. The question is whether companies are actually opening their wallets—and in this case, Ramp’s data suggests at least some are.
Ara Kharazian. lead economist at Ramp Economics Lab. said the June results show US firms appeared to be making direct payments to DeepSeek. rather than only running DeepSeek’s open-source models on their own infrastructure. In a social media post quoted by the South China Morning Post. Kharazian described the trend as “probably the biggest sign that companies are looking for cheaper alternatives to OpenAI and Anthropic. ” with some willing to use cheaper. Chinese models. He added that this could mean “sending US data back and forth from China-hosted servers.”.
That distinction—open-source use versus paying for a hosted service—sits at the center of the unease. It is one thing to run a model internally. It is another to route information through someone else’s servers, especially when the core pitch of corporate AI is that it can be deployed with control.
DeepSeek’s momentum still looks modest when measured against the biggest names. Kharazian said DeepSeek had a “modest hype cycle” in January 2025. when corporate adoption on Ramp rose to 0.3% before falling back to 0.1%. according to the South China Morning Post. By April 2026, DeepSeek’s adoption rate was still 0.1%.
In the same April comparison, Anthropic and OpenAI led the Ramp AI Index at 34.4% and 32.3%, respectively. The SCMP report did not provide June market share figures, but the direction is clear: DeepSeek is gaining attention from a small base, not taking over.
That small-base shift lands as enterprise AI budgets face more scrutiny. Many companies have moved beyond experimenting with chatbots and started embedding AI into coding, customer support, analytics, and operations. As those deployments expand, token costs, subscription tiers, and infrastructure demands can become harder to ignore.
And money may be coming to help DeepSeek push further. Proactive reported that DeepSeek is raising $7.4 billion in its first external funding round. The report put its valuation between $52 billion and $59 billion. It named Tencent, CATL, NetEase, and JD.com among the investors, and said founder Liang Wenfeng was also committing capital.
If the funding is finalized, it could strengthen DeepSeek’s ability to compete on infrastructure, product development, and enterprise sales—while reinforcing demand for lower-cost AI models outside the usual Silicon Valley vendor set.
For companies, the biggest practical challenge may not be model performance alone. It may be governance—what happens when cost-cutting leads to new vendor relationships, new hosting choices, and new questions about how data moves.
Kharazian’s comments suggest that some US firms may be sending data through DeepSeek’s hosted service rather than only using its open-source models internally. Ramp’s June list. then. reads less like a trend headline and more like an early signal that purchasing decisions in AI are getting more selective—driven by cost. but complicated by risk.
DeepSeek’s rise on Ramp’s list suggests enterprise AI buying is becoming more about trade-offs than brand loyalty. The winners, at least at first, may not always be the vendors with the biggest reputations. They may be the ones that help companies control costs—without turning vendor risk into a new problem to manage.
DeepSeek Ramp Silicon Valley AI costs enterprise AI OpenAI Anthropic data security vendor risk AI funding Liang Wenfeng Tencent CATL NetEase JD.com
So they’re using DeepSeek now? lol that seems risky.
I saw “sending data back and forth” and I’m like… why would any company sign off on that. If it’s cheaper, cool, but security should not be optional.
Wait is this saying they’re not running it themselves? I thought open source meant they could just download it and done. But apparently they’re paying DeepSeek directly which like… that’s totally different. Also Ramp tracking list sounds made up to me.
DeepSeek is trending on some spending app so that means the whole AI market is switching now right? /s Idk, I saw OpenAI costs went up and everyone panicked, but China hosted servers?? that’s gonna be a no from me even if the models are “cheaper.”