Nvidia wants to own your AI data center from end to end

"The implicit message of the imposing wall of racks is that Nvidia, if it doesn't already, will ultimately own all processing in the data center, from one end to the other. Existing product lines include the Vera CPU chip, the Rubin GPU chip, and, now, a new kind of rack of equipment joins them, for ultra-fast inference, called the LPX."

"The Groq 3 LPU "can combine the extreme FLOPS [floating-point operations per second] of GPUs and the bandwidth of LPUs into one," said Ian Buck, Nvidia's head of hyper-scale and high-performance computing. The LPX rack, which will be available later this year, is made up of chips Nvidia has designed using intellectual property it licensed in December from AI startup Groq for $20 billion."

"Nvidia argues that AI economics are better when all the parts are from Nvidia. Nvidia's broadening ambition includes robotics and even AI in space, demonstrating the company's expansion beyond traditional data center infrastructure into emerging application domains."

Nvidia showcased an expanded AI infrastructure ecosystem at its GTC conference, displaying five racks covering complete data center processing needs. The company announced new product lines including the Vera CPU chip, Rubin GPU chip, and the LPX rack designed specifically for ultra-fast AI inference. The LPX combines Groq 3 LPU chips, acquired through Nvidia's $20 billion licensing deal with AI startup Groq in December, with Rubin GPUs to balance inference speed and data handling capacity. Nvidia's strategy emphasizes that AI economics improve when all infrastructure components come from a single vendor. The company's ambitions extend beyond data centers into robotics and space-based AI applications.

Read at ZDNET