Microsoft CEO Describes DeepSeek as 'Real Innovation' and 'Good News'
Microsoft's CEO has praised DeepSeek for its 'real innovation' and described it as 'good news.'
TroibNews 
DeepSeek, the Chinese AI model that has astonished the tech sector and unsettled Wall Street due to its remarkably low cost and strong performance, has "some real innovation" and is "all good news," according to Satya Nadella, the CEO of Microsoft.
Nadella made this statement during a quarterly earnings call for the U.S. software giant on Wednesday, emphasizing that AI development cycles are "no different" from traditional computing.
On the same day, Microsoft rolled out the DeepSeek-R1 reasoning model, which allows users on its cloud platform to observe the AI’s "thought process."
Allegations Amid Use
Microsoft, a significant investor in OpenAI, DeepSeek's U.S. competitor, is currently investigating whether the Chinese AI developer has illicitly obtained data from OpenAI.
OpenAI has claimed to have evidence that DeepSeek utilized its services to train its AI models, a violation of OpenAI's terms of service.
However, neither OpenAI nor Microsoft has provided any supporting evidence for these claims.
While the investigation continues, a nominee from the Trump administration has already accused DeepSeek of employing "stolen" U.S. technology.
Howard Lutnick, Trump's pick for commerce secretary, informed U.S. senators on Wednesday that the government would take action regarding the matter.
Additionally, Trump’s AI adviser David Sacks mentioned to the media that there is "substantial evidence" indicating that DeepSeek "distilled the knowledge" from OpenAI's models.
"Distillation will violate most terms of service, yet it's ironic – or even hypocritical – that Big Tech is calling it out," remarked tech investor and Cornell University lecturer Lutz Finger in a statement on Wednesday. He further noted, "Training ChatGPT on Forbes or New York Times content also violated their terms of service."
Understanding 'Distillation'
Distillation refers to a method in which a new AI model queries a larger one repeatedly to learn from its outputs.
While mentioned in DeepSeek's publicly available research papers, the researchers stated that their implementation differed. They claimed that the DeepSeek-R1 reasoning model served as the "bigger one" to "teach" other models, such as Alibaba's Qwen and Meta's Llama, to enhance their reasoning capabilities.
DeepSeek has made both the distilled models and the original R1 available for free download, enabling users with less powerful computers—and even smartphones—to run these models offline with complete control. This contrasts with ChatGPT users, for whom the underlying model remains inaccessible, even for those who pay for access.
Social media discussions have suggested that DeepSeek sometimes identifies itself as ChatGPT, possibly indicating data theft. However, like other AI models, DeepSeek does not always provide accurate information. For instance, an earlier version of the Google Gemini chatbot identified itself as Baidu's Ernie bot when queried in Chinese, but Baidu never accused Google of data theft.
Rohan Mehta contributed to this report for TROIB News
Nadella made this statement during a quarterly earnings call for the U.S. software giant on Wednesday, emphasizing that AI development cycles are "no different" from traditional computing.
On the same day, Microsoft rolled out the DeepSeek-R1 reasoning model, which allows users on its cloud platform to observe the AI’s "thought process."
Allegations Amid Use
Microsoft, a significant investor in OpenAI, DeepSeek's U.S. competitor, is currently investigating whether the Chinese AI developer has illicitly obtained data from OpenAI.
OpenAI has claimed to have evidence that DeepSeek utilized its services to train its AI models, a violation of OpenAI's terms of service.
However, neither OpenAI nor Microsoft has provided any supporting evidence for these claims.
While the investigation continues, a nominee from the Trump administration has already accused DeepSeek of employing "stolen" U.S. technology.
Howard Lutnick, Trump's pick for commerce secretary, informed U.S. senators on Wednesday that the government would take action regarding the matter.
Additionally, Trump’s AI adviser David Sacks mentioned to the media that there is "substantial evidence" indicating that DeepSeek "distilled the knowledge" from OpenAI's models.
"Distillation will violate most terms of service, yet it's ironic – or even hypocritical – that Big Tech is calling it out," remarked tech investor and Cornell University lecturer Lutz Finger in a statement on Wednesday. He further noted, "Training ChatGPT on Forbes or New York Times content also violated their terms of service."
Understanding 'Distillation'
Distillation refers to a method in which a new AI model queries a larger one repeatedly to learn from its outputs.
While mentioned in DeepSeek's publicly available research papers, the researchers stated that their implementation differed. They claimed that the DeepSeek-R1 reasoning model served as the "bigger one" to "teach" other models, such as Alibaba's Qwen and Meta's Llama, to enhance their reasoning capabilities.
DeepSeek has made both the distilled models and the original R1 available for free download, enabling users with less powerful computers—and even smartphones—to run these models offline with complete control. This contrasts with ChatGPT users, for whom the underlying model remains inaccessible, even for those who pay for access.
Social media discussions have suggested that DeepSeek sometimes identifies itself as ChatGPT, possibly indicating data theft. However, like other AI models, DeepSeek does not always provide accurate information. For instance, an earlier version of the Google Gemini chatbot identified itself as Baidu's Ernie bot when queried in Chinese, but Baidu never accused Google of data theft.
Rohan Mehta contributed to this report for TROIB News
