DeepSeek gets Silicon Valley talking

Starting from Chinese AI company DeepSeek released an open source version of his R1 concept Earlier this week, many in the tech industry have been talking a lot about the company’s findings, and what it means for AI.

for example, the capitalist Marc Andreessen, has been sent DeepSeek is “one of the most amazing and fascinating things I’ve ever seen.”

R1 seems to match or beat the OpenAI version of o1 on some AI benchmarks. And the company claims one of its products it only cost $5.6 million training, compared to the hundreds of millions of dollars that America’s leading companies pay to train theirs.

It also appears to have achieved this in the face of US sanctions that restrict the sale of high-end chips to Chinese companies. MIT Technology Review writes that the company’s success shows how sanctions are “driving startups like DeepSeek to innovate in ways that prioritize efficiency, resource utilization, and collaboration.” (On the other hand, The Wall Street Journal reports (that DeepSeek’s Liang Wenfeng recently told the Chinese Premier that America’s export restrictions are still a hindrance.)

Curai Chief Neal Khosla he gave a simple explanationsaying that the company is a “ccp state psyop” which “says that the price was low which means low price and hope that everyone will change (to) destroy the AI ​​competition in us.” (A Community Notice is attached to his post indicating that Khosla does not testify to this, and that his father Vinod is an OpenAI investor.)

Meanwhile, journalist Holger Zschaepitz reported DeepSeek can “represent a big threat to the US market” – if a Chinese company can make a cheap model at a low price, without getting many chips, it can doubt “spend hundreds of billions of dollars. capex is pouring into these companies.”

In response, Y Combinator CEO Garry Tan disagreed The success of DeepSeek may be good for the American competitors: “If the training models become cheaper quickly and easily, the need for knowledge (the real use of AI in the world) will grow and become faster, which ensures that the computing will be used.”

And Chief Meta AI Scientist Yann LeCun they argued looking at DeepSeek’s announcement through the lens of China vs. United States. In fact, he said the real lesson is that “open brands outperform their owners.”

“DeepSeek has benefited from open research and open source (eg PyTorch and Llama from Meta),” LeCun wrote. “He came up with new ideas and built them on top of other people’s work. Because their work is published and open source, anyone can benefit from it. “

All the controversy seems to be driving consumers to try the products – starting Sunday afternoon, DeepSeek’s AI assistant is the top free app in the Apple App Store, ahead of ChatGPT.