Tanner Brown
While major tech players in the West evaluate the level of disruption to come from the breakout performance of Chinese AI platform DeepSeek, with its supposed high performance and low cost, a different narrative is unraveling in China—among the government, media, rival tech firms, and the country’s rapidly expanding AI user base.
“It’s thrilling,” said Cai Minghan, a Taiwan-based tech analyst. He told Barron’s the Chinese AI scene had been set into a frenzy, as in the U.S., but that the nature of it was different.
The start-up claimed its new version matches global leader OpenAI’s best model at a fraction of the cost and without access to the most advanced semiconductor chips. Based in Hangzhou, China, where Alibaba Group Holding has its headquarters, DeepSeek also made its models open source, allowing anyone to harness and alter the product for their own use.
“Competitors are racing to show that they also have cutting-edge models, but their stocks have risen, and their forecasts have brightened,” he said. That elation is in large part due to the as-yet unproven assertion that DeepSeek used less advanced Nvidia chips and spent a mere $6 million in its final training round. Given that, and the opportunity to harness DeepSeek’s open-source framework, Chinese firms’ outlooks have only improved, Cai said.
No comments:
Post a Comment