Intel's former CEO has used DeepSeek instead of OpenAI in his startup Gloo

Odaily reported that DeepSeek's new open-source AI inference model R1 has triggered a sell-off of Nvidia's stock and made its consumer application top the Apple App Store chart. Last month, DeepSeek stated that it used a data center with about 2,000 Nvidia H800 GPUs and trained a model within two months at a cost of around $5.5 million. Last week, DeepSeek published a paper showing that the performance of its latest model matches the world's most advanced inference models. These models are being trained in data centers that have spent billions of dollars on purchasing Nvidia's faster and very expensive AI chips. The entire tech industry has reacted very strongly to DeepSeek's high-performance, low-cost model. For example, former Intel CEO Pat Gelsinger posted on saying: "Thank you, DeepSeek team." Gelsinger recently stepped down as Intel's CEO and is now the chairman of his own IPO startup Gloo. He worked at Intel for four years and tried to catch up with Nvidia using Intel's alternative AI GPU Gaudi 3 AI, but left the company in December. He wrote that DeepSeek should remind the tech industry of the three most important lessons: lower cost means wider adoption; creativity thrives under constraints; and open source is the way to win. DeepSeek will help reset the increasingly closed world of foundational AI model work. Gelsinger also revealed that R1 is so impressive that Gloo has decided to no longer use and pay for OpenAI. Gloo is building an AI service called Kallm, which will provide chatbots and other services.