China's Open-Source AI Models Might Be Outpacing American Companies In Cost
While OpenAI's ChatGPT, Microsoft's CoPilot, and Google's Gemini are the talk of the town, those aren't the AI models in which many businesses are investing. Lesser-known Chinese AI models like DeepSeek and Alibaba have seen a nearly 30% hike in usage from businesses worldwide since the end of 2024. The one feature that sets these AI models apart from the big American names in the industry is cost. They're either cheaper than adopting OpenAI's model or virtually free, which successful businesses are always fond of.
Some U.S. businesses have been able to save as much as $400,000 in a year by using these alternative models. Typically, premium AI models like ChatGPT are seen as superior, especially to something that costs next to nothing, but even the CEO of Airbnb, Brian Chesky, admitted to Bloomberg that his company chose not to integrate OpenAI's model initially for its chatbot, saying, "I didn't think it was quite ready."
However, while Airbnb does utilize OpenAI a bit with its chatbot, Chesky admitted his company leans more on other AI models: "We're relying a lot on Alibaba's Qwen model. It's very good. It's also fast and cheap. We use OpenAI's latest models, but we typically don't use them that much in production because there are faster and cheaper models." Investors have poured billions of dollars into companies like OpenAI and Anthropic, only for these firms' very customers to turn to their inexpensive, overseas competitors, spelling trouble for the American-made models.
It's not just the low cost drawing in businesses
Beyond being cheaper for companies to integrate, Chinese-made AI models are also typically open-sourced, giving companies more freedom to do with them what they want. AI models like ChatGPT and Gemini tend to be proprietary, preventing customers from modifying the code at all. Once upon a time, this wasn't the case. Several major companies, such as Bloomberg, attempted to create their own tools that used open-sourced AI models, but they were lightyears behind ChatGPT's capabilities.
Fast forward to 2026, and that's no longer the case. These open-sourced models have quickly closed the gap, letting users utilize AI tools any weird way they want. It's not just corporations gravitating to the Chinese models, either. According to NBC News, Jerry Liu, founder of DayFlow, estimates that around 40% of its users opt for an open-source, Chinese-made, models over a closed American-made one. This leap in functionality could be because the developers of these open-source models copied much of what OpenAI and Anthropic created.
When you consider that DeepSeek and Alibaba are both cheaper and offer more flexibility for nearly the same amount of capability, it should come as no surprise that they're more enticing than their American counterparts. There are risks, however, to using platforms like DeepSeek and Qwen. So much so that the U.S. Navy banned the use of DeepSeek because of privacy concerns.