DeepSeek stunned the AI industry when it unveiled its flagship model R1 earlier this year.
Microsoft CEO Satya Nadella said R1 was the first model he had seen that could match up to OpenAI’s.
OpenAI CEO Sam Altman called R1 an “impressive model” in January.
Microsoft CEO Satya Nadella said the Chinese AI startup DeepSeek’s R1 model was the first AI model he had seen that was as competitive as OpenAI’s.
“OpenAI has been so far ahead that no one’s really come close,” Nadella said in an interview with Bloomberg Businessweek published Thursday.
“DeepSeek, and R1 in particular, was the first model I’ve seen post some points,” Nadella added.
In January, DeepSeek spooked investors after its free chatbot app hit the top spot on Apple’s US App Store’s ranking. DeepSeek’s high-performing but relatively cheaper models sparked a sell-off in AI-related stocks as investors questioned if it would cause demand for AI hardware like Nvidia’s chips to fall.
Microsoft started offering versions of DeepSeek’s R1 model on its cloud platform, Azure AI Foundry, in January. Aside from DeepSeek, the platform includes other AI models from companies like OpenAI, Meta, and Mistral.
Using R1 on Microsoft’s platform meant that data would not be sent to DeepSeek’s servers in China.
Asha Sharma, a Microsoft corporate vice president who heads the company’s AI platform product, wrote in a blog post in January that R1 went through “rigorous red teaming and safety evaluations” before it was made available to customers.
In January, OpenAI CEO Sam Altman called R1 “an impressive model, particularly around what they’re able to deliver for the price.” Altman said OpenAI would “pull up some releases” in response to DeepSeek’s “invigorating” competition.
“R1 is actually not that unusual,” Ben Buchanan, a former special advisor for artificial intelligence in the Biden administration, said in an episode of “The Ezra Klein Show” which aired in March.
Buchanan said while DeepSeek’s engineers are “extremely talented,” he didn’t think the “media hype around it was warranted.”
“Where do you think they got their performance increases from? We read their papers. They’re smart people who are doing exactly the same kind of algorithmic efficiency work that companies like Google and Anthropic and OpenAI are doing,” Buchanan added.
Microsoft, OpenAI, and DeepSeek did not respond to requests for comment from Business Insider.
Read the original article on Business Insider
