Aakash Gupta, a tech entrepreneur, has observed that most startups pitching to venture capital firm Andreessen Horowitz are opting for low-cost Chinese open-source AI models over leading U.S. systems. He noted that this cost gap is significantly influencing early-stage product and funding strategies. Gupta made this statement on the social media platform X.
"80% of startups pitching Andreessen Horowitz are running on Chinese open-source models," said Gupta, Newsletter Author and Podcast Host. "Not OpenAI. Not Anthropic. Chinese models like DeepSeek that cost 214x less per token."
According to Bloomberg, these low-cost Chinese open-source AI models are gaining interest from global developers and affecting adoption patterns in Silicon Valley. The publication highlights that these models allow for broader experimentation at reduced costs. Fortune reports that China's focus on accessible open-source systems is reshaping the competitive landscape for foundational AI technologies.
Reuters states that DeepSeek reported training its R1 AI model for $294,000, a figure considerably lower than the costs associated with major U.S. frontier systems. The company emphasized that minimizing training expenses was a central component of its strategy. This disclosure has intensified market scrutiny regarding the cost structure of Chinese AI development.
The Wire China explains that affordable open and open-weight Chinese AI models are rapidly being adopted by global startups. These systems provide companies with the ability to experiment with AI infrastructure at significantly lower operating costs, thereby expanding access to advanced model capabilities in markets traditionally limited by budget constraints.
Gupta is also known as the author of "Product Growth," a newsletter and podcast centered on AI and product management. His professional background includes leadership roles at Apollo.io, Affirm, Epic Games, and thredUP. Gupta holds an MBA from the Wharton School and a bachelor's degree from the University of Michigan.
