Since Chinese AI company Deepseek released an open source version of its R1 reasoning model earlier this week, many in the tech industry have been making big pronouncements about what the company achieved, and what it means for the state of AI. .
Venture capitalist Marc Andreessen, for example, tweeted that Deepseek is “one of the most amazing and impressive advances I’ve ever seen.”
The R1 seemingly matches or beats Openai’s O1 model on several AI benchmarks. And the company claims that one of its models cost just $5.6 million to train, compared to the hundreds of millions of dollars that major US companies pay to train theirs.
It also appears to have succeeded in the face of US sanctions banning the sale of advanced chips to Chinese companies. The MIT Technology Review writes that the company’s success illustrates how sanctions are “driving startups like Deepseek to innovate in ways that prioritize efficiency, resource pooling, and collaboration.” (On the other hand, the Wall Street Journal reports that Deepseek’s Liang Wenfeng recently told China’s Premier that US export restrictions still pose an obstacle.)
Curai CEO Neal Khosla offered a simpler explanation, claiming the company is a “state CCP PSYOP” that is “falsifying the cost was low to justify setting the price low and hoping that everyone switches to to (harm) AI competition in the US” (A community note is attached to his post noting that Khosla offers no evidence for this, and that his father Vinod is an open investor.)
Meanwhile, journalist Holger Zschaepitz suggested Deepseek “could represent the biggest threat to US capital markets”—if a Chinese company could build an advanced model at low cost, without access to advanced chips, it would come in doubt “the utility of the utility of the Hundreds of billions worth of the hat that are poured into this industry. “
In response, Y Combinator CEO Garry Tan argued that Deepseek’s success would bode well for its American competitors. “If training models become cheaper and easier,” he wrote in X, “the demand for inference (the actual real-world use of AI) will grow and accelerate even faster, which ensures that the supply of computation will to be used.”
And Meta’s Chief Scientist Yann Lecun argued for looking at Deepseek’s announcement through the lens of China versus the United States. Instead, he suggested the real lesson is that “open source models are outpacing proprietary ones.”
“Deepseek has benefited from open research and open source (eg Pytorch and Llama from Meta),” Lecun wrote on LinkedIn this week. “They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can benefit from it.”
All the controversy seems to be pushing consumers to try the product. As of Sunday afternoon, Deepseek’s AI assistant is the top free app in the Apple App Store, just ahead of Chatgpt.
