Tesla CEO Elon Musk at the 2017 session of the neural information processing system, said Tesla is actively from the software, hardware to develop artificial intelligence technology, and revealed that Tesla has been put into the development of artificial intelligence chip. If Musk can do what it wants, it is likely to pose a big threat to Nvidia's market position. According to the Motley Fool website, nvidia first stepped into the AI market and took the initiative. Nvidia is adept at parallel processing of GPU, which is suitable for the training of artificial intelligence systems. NVIDIA's data center revenue has grown 600% in two years, and the company's overall revenue has been doubled. Yet Tesla's attempt to develop artificial intelligence may not make nvidia too long. Tesla in 2016 to the former Apple and Micro (AMD) chip engineer Jim Keller. Keller for Apple, is responsible for the development of customer-made chip. Musk said that Keller for Tesla to create a customized chip, will be the world's best artificial intelligence dedicated hardware. On the other hand, the micro-is constantly trying to improve its own competitiveness in the field of artificial intelligence. Earlier, Tesla has been working with the micro-development of AI chip, but the news has not been confirmed by both sides. In any case, these vendors do not want NVIDIA to continue to enjoy the advantages of a pioneer. Alphabet's Google is one of the pioneers in the development of artificial intelligence. The TPU (Tensor 處理 unit), published in 2016, has been in the Google data Center for more than a year. In addition, Google is also one of the long-term customers of the NVIDIA GPU. In early 2017, Google launched the second generation of TPU, which was able to perform training and inference (inference) at the same time. In the past, Google TPU can only engage in artificial intelligence inference, so the algorithm training relies on the Nvidia GPU. After the introduction of the second generation of TPU, Google may be able to escape the influence of the NVIDIA GPU. Intel is putting the artificial intelligence campaign on two fronts. Intel first cost 16.7 billion dollars to buy Altera, achieved FPGA technology, and in 2016 years spent about 400 million U.S. dollars to acquire new company Nervana, the development of artificial intelligence ASIC chip. Nervana attempts to remove components that are not required for AI applications from the GPU and claims that this will allow the GPU to increase its computing power to 10 times times the current size. The development of artificial intelligence is still in the initial stage. While Nvidia has a few advantages, it doesn't mean that the GPU will always be the champion. Tesla is developing AI chips and is just one of the many challengers that Nvidia will face.
