The emergence of Chinese AI startup DeepSeek has taken the world by storm, especially the company released a research paper claiming that its R1 model surpasses proprietary AI models like OpenAI’s o3 model across a wide range of benchmarks at a fraction of the development cost.
Shortly after the launch, reports claimed that DeepSeek unlawfully used OpenAI’s data to train its cost-effective model. The ChatGPT maker’s CEO, Sam Altman, admitted that DeepSeek’s R1 AI model was quite impressive but claimed OpenAI was better positioned to “obviously deliver much better models” as it strives to hit the coveted AGI benchmark.
On the other hand, Microsoft CEO Satya Nadella touted DeepSeek’s AI model as “super impressive” and urged key stakeholders in the AI landscape to “take the developments out of China very, very seriously.”
DeepSeek’s cost-effective and open-source AI model became a significant concern for Microsoft, prompting Satya Nadella to set up a virtual war room using Microsoft Teams to investigate the Chinese startup’s overnight success.
While speaking to Bloomberg Businessweek, Satya Nadella indicated:
“OpenAI has been so far ahead that no one’s really come close. DeepSeek, and R1 in particular, was the first model I’ve seen post some points.”
Over 100 Microsoft employees were tasked to scrutinize DeepSeek’s research paper and access the models security. “People didn’t sleep,” indicated Asha Sharma, Microsoft’s AI platform lead.
And after approximately 48 hours, the details shared in DeepSeek’s research paper seemingly checked out. In a perfect world, a company would quickly jump back to the drawing board in an attempt to compete with the new kid in the AI block, but not Microsoft.
Satya Nadella reportedly instructed his team to install DeepSeek’s R1 AI model on Microsoft’s cloud, selling its access to customers alongside other products from competitors in the space, including OpenAI. “Get it out,” Nadella recounts the moments leading to R1 being installed in Microsoft’s cloud.
As such, the situation sheds light on Microsoft’s multi-billion-dollar partnership with OpenAI, potentially indicating that the software giant’s allegiance isn’t to OpenAI’s exorbitant AI models and technology. Instead, Satya Nadella is more focused on making whatever AI models or products accessible to customers via Microsoft’s cloud.
Regardless of the AI model or service a client chooses from Microsoft’s broad portfolio in Azure, the software giant will still be compensated for its cybersecurity, cloud computing, and data storage services.
