In 5 of the GPU (graphics processor) technology conference, huang, chief executive of nvidia announced the latest dedicated to Tesla P100 graphics processing chip in the field of artificial intelligence research, he claims to be the development of the GPU, nvidia has spent $2 billion in research and development costs.

What concept is $2 billion? Nvidia's revenue for all of last year was $5 billion, so the nvidia spent almost half of the year income into the chip research and development, the reasoning behind this is that nvidia took a fancy to the development of artificial intelligence in the future.

Nvidia bullish on artificial intelligence (ai) $2 billion chip

Launched over the same period of the chip are also applied to deep learning model for DGX - 1 a supercomputer, nvidia said the $129000 computer, equivalent to integrated with 250 server, has seven terabytes of SSD storage, 8 Tesla P100 released the same day the GPU and two Xeon processor, the whole equipment is last year's published 12 times super computing ability.

Executive vice President of Microsoft's global shen xiangyang recently said in an interview with tencent technology such as media, with the rapid development of artificial intelligence, make such as nvidia is engaged in production of hardware development company is very excited.

"You see last GDC (global game developers conference), nvidia's CEO jen-hsun huang do speak completely in the theme of machine learning, it is unthinkable, a company that do the GPU is machine learning." Shen xiangyang said.

Shen xiangyang believe that artificial intelligence research field in recent years, the outbreak of the reason lies in three aspects, namely, large data, large calculation, coupled with precise algorithm. Among them, the large computing and hardware processing power are inseparable.

The researchers found that in recent years in the field of artificial intelligence in the processing of large amounts of data for image recognition, speech recognition and other deep learning related work, compared to traditional GPU and CPU has a unique advantage. In the past in terms of deep learning, GPU computing power is usually 10 to 20 times the CPU. Even if Intel and the computing power is the strongest Xeon processor, nvidia said, its focus on Tesla gpus in the field of artificial intelligence computing power is 3 to 5 times more.

With the rapid development of artificial intelligence, nvidia has also realized the importance of a specific market, in his speech, huang said that in the next 10 years, artificial intelligence market value of about $500 billion.

Under such a judgment, nvidia that needs to be developed specifically for this market targeted GPU products, stage, in a speech yesterday said jen-hsun huang, nvidia "designed and developed specifically for the first time for artificial intelligence computing power and accelerate the depth study of image processing chip architecture", a $2 billion development costs are included.

At present, including Google, Facebook, Microsoft and other technology companies, the leader of the field of artificial intelligence research, and has been provided by the use of nvidia chips specially applied in the field of research.

Facebook's artificial intelligence laboratory director Yann LeCun, said: "the nvidia gpus are accelerating the development of artificial intelligence. Getting larger as the neural network, we need not only the GPU memory is bigger, faster, also need to be improved GPU speed of communication between the operations and able to take advantage of lower precision hardware."