Former Intel CEO, Pat Gelsinger, has opted for DeepSeek in lieu of OpenAI for running operations at his startup, Gloo. This decision reflects the incredible performance of DeepSeek’s new open source AI reasoning model, R1.
DeepSeek made headlines last month when it revealed the training of its model utilizing a cluster of Nvidia’s H800 GPUs. This process, taking only two months, cost a ballpark figure of $5.5 million. The subsequent release of a report demonstrated R1’s performance is at par with the top reasoning models in the world.
In response to DeepSeek’s achievement of high efficiency at reduced cost, Gelsinger, a seasoned hardware engineer and the incumbent chairman of his own startup, Gloo, expressed his admiration publicly.
As per Gelsinger, the lower costs of DeepSeek’s model will play a key role in its widespread adoption. He was particularly excited about the open source nature of DeepSeek, contrasting sharply with the often closed environments of foundational AI model work. “Open Wins. DeepSeek will help reset the increasingly closed world of foundational AI model work”, he said.
This influenced tremendously Gloo’s decision not to opt for OpenAI. The startup plans to develop an AI service, Kallm, offering a chatbot and other services. While the rest of the tech world continues to speculate on DeepSeek’s performance and cost, Gelsinger has full confidence in DeepSeek, attributing its progress to engineering creativity.
Original source: Read the full article on TechCrunch
