TapTechNews June 3rd news, Nvidia's chief executive, Jensen Huang, recently put forward an interesting concept - CEO Math on the eve of Computex Taipei 2024.
The more you buy, the more you save, Huang said in his speech. This is CEO Math, which is not entirely accurate but very effective.
Puzzling at first listen? Jensen Huang then explained the meaning of this concept. He suggested that enterprises invest in both graphic processing units (GPUs) and central processing units (CPUs) simultaneously. These two types of processors can work together to shorten the task completion time from 100 units to 1 unit. Therefore, in the long run, increasing investment can actually save costs.
This practice of combining the use of CPUs and GPUs is already very common in the personal computer field. We add a $500 GPU to a $1,000 computer, and the performance will be greatly improved, Jensen Huang gave an example. The same principle also applies to data centers. Adding a $500 million GPU to a data center worth $1 billion, it suddenly turns into an AI factory.
Jensen Huang also showed a chart that shows that when enterprises use CPUs and GPUs simultaneously, the speed can be increased by 100 times, while the cost only increases by 1.5 times.
In March this year, Nvidia released the so-called world's most powerful AI chip - Blackwell B200 GPU, with a price tag of up to $70,000 (TapTechNews note: Currently about 508,000 RMB). Nvidia also integrated this chip into a larger-scale design, such as GB200 NVL72, which combines 72 GPUs and 36 CPUs, dedicated to the most computationally intensive workloads, and can reduce costs and energy consumption by up to 25 times.
In recent years, Nvidia has frequently made headlines as a key player in the AI wave. In the fourth quarter of 2023, Nvidia's revenue exceeded 22 billion US dollars.