Original | Text by BFT Robot
1. The competition among AI chips has become increasingly fierce.
Although various AI chips with different functions and positioning can complement each other to a certain extent, they are also constantly adjusting their positioning amid the coexistence of opportunities and challenges. On the AI training side, Nvidia's GPU has always been the first choice for training due to its high computing power threshold.
Only a few chips can rival him, such as Google's TPU (which has training and inference capabilities since 2017). In addition, when the algorithm begins to stabilize and mature, ASIC customized chips can also take on part of the computing power to a certain extent due to their specificity and low energy consumption. In comparison, the inference end has lower computing power requirements than the training end, and the market demand is larger and more segmented. In addition to the current mainstream CPU, chips such as GPU, FPGA and ASIC can also occupy a seat. Nvidia also launched a variety of inference chips in this GTC, and hopes to replicate its success on the training side in this market.
The future of the AI inference market is promising, and Nvidia continues to increase its efforts to regain the dominant position of CPUs in the inference end. Although the inference end is currently dominated by CPUs, supplemented by other chips such as older GPUs and ASICs, in the long run their GPUs should have greater advantages in the inference end.
According to news reports, China plans to create a new science and technology policy, accelerate the construction of national computing infrastructure, and lay the foundation for the development of artificial intelligence. The Ministry of Industry and Information Technology of China has expressed support for the development of generative AI models such as Alibaba and Baidu. In response to the United States' suppression of my country's chip technology, Shanghai Biren, a start-up company, openly seeks to accelerate its first public offering in Hong Kong, aiming to replace Nvidia's chip position.
The AI chip competition landscape is intensifying. In addition to chip companies in the competition circle, other Internet companies have also joined this wave of competition, such as foreign Google, Tesla, Amazon, etc. Our country does not want to fall behind. Baidu is early Entering the game to build cores, Kuaishou, Meituan, Alibaba, etc. also joined the battle. As a result, Nvidia's position as the king in the artificial intelligence world will be somewhat affected. How far can the exploding "Nvidia" era go in this "core war"?
2. NVIDIA announced that it will join forces with two major Indian companies to develop large supercomputing and AI models.
In order to consolidate its leading position in the AI industry, NVIDIA has cooperated with Reliance Industries Group, India's largest private enterprise. It is said that it will set up supercomputing facilities in India and jointly build local large-scale models to make AI technology better for 14 Although this factor may exist, it is by no means the key factor. NVIDIA chose Reliance Industries, firstly, because Reliance Industries has a huge user base and data resources, and secondly, because Reliance Industries has strong political influence and social responsibility.
Nvidia and Reliance Industries will build an AI infrastructure that will be an order of magnitude more powerful than the fastest supercomputers in India today. NVIDIA will provide the most advanced GH200 Grace Hopper super chip and cloud AI super computing service DGX Cloud.
Then, on Friday, Nvidia announced that it would cooperate with the Tata Group, India's largest group company, saying that it would provide AI computing infrastructure and platforms for developing AI solutions, and the two companies would jointly build an An AI supercomputer powered by the next generation NVIDIA GH200 Grace Hopper super chip.
Nvidia specifically mentioned cooperation with two companies under the Tata Group. Among them, Tata Communications will develop an AI cloud with Nvidia, aiming to provide key infrastructure to support the next life cycle of computing.
Tata Consultancy Services will use AI infrastructure and capabilities to build and process generative AI applications. With the cooperation with NVIDIA, the company will restructure its cooperation with customers through an AI-first approach and will also improve 60 skills of thousands of employees.
Nvidia said that the cooperation will also promote AI-driven transformation within the entire Tata Group, including manufacturing and consumer businesses. As the world's leading graphics card manufacturer, NVIDIA hopes to better understand and adapt to the needs of the local market by cooperating with local companies. In addition, entering new markets will help expand the company's business scope and further consolidate its leading position in the GPU market.
In the context of globalization, all countries are seeking a more independent and controllable path to scientific and technological development. This cooperation is also a way for Nvidia to support local industries in India. By working with local partners to jointly develop large AI models and supercomputing systems suitable for the local market, it will help promote scientific and technological exchanges and cooperation between India and NVIDIA, benefiting more Indian companies and people.
AI large models are also a current research hotspot in the field of artificial intelligence. By cooperating with Indian, NVIDIA can share its research and application experience in large AI models with local partners to further promote the development of AI technology on a global scale.
This kind of cooperation will not only help improve local technological levels, but also help promote the development of the global AI industry and bring more scientific and technological progress to human society. It will also help increase NVIDIA's exposure in the Indian market and further expand the company's business scale in this market.
Author | Ju Jushou
Typography | Spring Flowers
Review | Cat
If you have any questions about the content of this article, please contact us and we will respond promptly. For more information, please pay attention to BFT intelligent robot system~