With the explosion of AI applications such as ChatGPT and Wenxin Yiyan, the demand for computing power to support artificial intelligence has also skyrocketed. Baidu said that it will release the generative AI product "Wen Xin Yi Yan" on March 16. In order to support Wenxin Yiyan's ultra-large-scale computing needs and further realize the industrialization of Wenxin Yiyan, Baidu Smart Cloud has also frequently announced the preparations for Wenxin Yiyan's supporting facilities recently.
On March 14, Baidu Smart Cloud stated that since the end of last year, Baidu Smart Cloud has comprehensively upgraded its cloud service capabilities through three major actions: In December last year, it released the first domestic full-stack self-developed AI infrastructure "AI Big Base" , In February this year, the AI R&D and Operation Integration (MLOps) capability was upgraded, and in March, Baidu Yangquan Intelligent Computing Center completed the upgrade. According to reports, the computing power infrastructure behind "Wen Xin Yi Yan" is supported by Baidu Intelligent Computing Center, and Baidu's multiple intelligent computing centers will also provide the underlying support for the large-scale implementation of "Wen Xin Yi Yan" for the industry.
The implementation of the generative AI industry faces three major challenges, and the rules of the cloud market are changing
Behind the upsurge of generative AI, the value of cloud computing to enterprises has undergone a fundamental qualitative change: from providing infrastructure to providing intelligent services, which also consolidates more possibilities for the combination of generative AI products and industries. Robin Li, founder and CEO of Baidu, said that "Wen Xin Yi Yan" is a large-scale model based on Baidu's intelligent cloud technology, which will fundamentally change the rules of the game in the cloud market.
And this subversive change also means that generative AI products such as Wenxinyiyan face multiple challenges in the industrial scale implementation. Specifically, the challenges are mainly reflected in the following three aspects: First, intelligent computing power. The data volume of generative AI products is huge, and the operation requires unprecedentedly powerful AI computing power, which poses a challenge to the existing computer architecture. The second is AI infrastructure. Artificial intelligence products such as Wenxin Yiyan have complex computing scenarios and multi-dimensional computing architectures. Enterprises need a full-stack AI infrastructure that integrates chips, frameworks, models, and applications for AI development, and end-to-end intelligence. The whole process of closed-loop AI development. The last is the level of AI engineering. When AI generative products are put into industrial practice, they need to carry out secondary adaptation development based on specific business scenarios. However, the entire process of AI model development and operation is highly complex, and irregular operations in any link will affect the effect of the AI model. The level of AI engineering determines whether AI can break through the "last mile" of industrial implementation.
Facing the new rules of the game in the cloud market, only cloud vendors that solve the above three challenges for enterprises are eligible to get admission tickets. Shen Dou, executive vice president of Baidu Group and president of Baidu Smart Cloud Business Group, said that cloud services have transitioned from the digital age to the smart age. In the past, enterprises chose cloud vendors more based on basic cloud services such as computing power and storage. Demand will focus more on intelligent service levels.
Wenxin Yiyan will be released soon, and Baidu Smart Cloud will upgrade three supporting cloud services
In the era of intelligent computing, enterprises are facing many challenges such as the above-mentioned shortage of intelligent computing power, the complexity of the whole process of AI development, and the difficulty of engineering the AI industry. Baidu said that it will provide the industry with three major core competencies to help the industry's intelligent transformation on the fast track.
First, it provides enterprises with a huge amount of high-performance intelligent computing power to support the industrial implementation of generative AI products such as Wenxin Yiyan. Baidu has currently built smart computing centers in Yangquan, Shanxi, and Yancheng, Jiangsu. Among them, Baidu Yangquan Intelligent Computing Center is the largest single intelligent computing center in Asia, with a construction scale of 4 EFLOPS (40 billion floating-point operations per second) AI computing power, which can meet the ultra-large-scale AI computing needs of various industries. Song Chunxiao, head of strategy at Kunlun Core Technology, said: "Artificial intelligence chips are the core of computing power. The second generation of Kunlun Core has been widely used in the application of Baidu Wenxin large models, and provides AI computing power for the intelligent upgrade of all walks of life. support."
It is understood that Baidu Intelligent Computing Center continues to innovate the computing architecture, supports the computing tasks of different business scenarios of enterprises in the era of intelligent computing, and improves the business efficiency and innovation capabilities of enterprises. At present, Baidu Intelligent Computing Center has supported cutting-edge AI applications such as Wenxin pre-trained large models, biological computing, and autonomous driving. At the same time, based on the self-developed innovative technology, the PUE can be as low as 1.08, realizing efficient and energy-saving operation, thereby reducing customers' electricity bills and operation and maintenance costs.
(Baidu Yangquan Intelligent Computing Center real picture)
Second, provide enterprises with new AI infrastructure to reduce costs and increase efficiency in AI development. Baidu Smart Cloud launched the "Baidu AI Base" by encapsulating computing power, framework, models, and AI applications. From the high-end chip Kunlun core, to the deep learning framework of the flying paddle, to the Wenxin pre-trained large model, and then to the AI application, an end-to-end intelligent closed-loop is realized, thereby turning the high-threshold AI technology into a power supply for enterprises like hydropower On-demand access, greatly reducing enterprise development costs and improving efficiency. Song Fei, General Manager of Baidu Intelligent Cloud Cloud Computing Product Solutions and Operations Department, said: "Baidu's AI base can be continuously optimized based on actual business data, so that the resource utilization rate can be increased to 70%, and the enterprise development efficiency can be increased by 100%."
Finally, it provides enterprises with AI R&D and operation integration (MLOps) capabilities, and accelerates the rapid industrial landing of generative AI and other large-scale model products. Xin Zhou, director of Baidu AI Central Taiwan, said that the Baidu AI base will provide a series of AI R&D and operation and maintenance tools for enterprises. After accessing Wenxin Yiyan, enterprises can easily and cost-effectively complete the adaptation and secondary development of business scenarios, and use AI engineering capabilities to help the industry break through the "last mile" of AI implementation.
With the support of low-cost multi-architecture intelligent computing power, "AI large base" new AI infrastructure, and MLOps AI engineering capabilities, Wenxin Yiyan will accelerate the arrival of industrial intelligence and bring true AI inclusiveness.
On March 14, Baidu Smart Cloud stated that since the end of last year, Baidu Smart Cloud has comprehensively upgraded its cloud service capabilities through three major actions: In December last year, it released the first domestic full-stack self-developed AI infrastructure "AI Big Base" , In February this year, the AI R&D and Operation Integration (MLOps) capability was upgraded, and in March, Baidu Yangquan Intelligent Computing Center completed the upgrade. According to reports, the computing power infrastructure behind "Wen Xin Yi Yan" is supported by Baidu Intelligent Computing Center, and Baidu's multiple intelligent computing centers will also provide the underlying support for the large-scale implementation of "Wen Xin Yi Yan" for the industry.
The implementation of the generative AI industry faces three major challenges, and the rules of the cloud market are changing
Behind the upsurge of generative AI, the value of cloud computing to enterprises has undergone a fundamental qualitative change: from providing infrastructure to providing intelligent services, which also consolidates more possibilities for the combination of generative AI products and industries. Robin Li, founder and CEO of Baidu, said that "Wen Xin Yi Yan" is a large-scale model based on Baidu's intelligent cloud technology, which will fundamentally change the rules of the game in the cloud market.
And this subversive change also means that generative AI products such as Wenxinyiyan face multiple challenges in the industrial scale implementation. Specifically, the challenges are mainly reflected in the following three aspects: First, intelligent computing power. The data volume of generative AI products is huge, and the operation requires unprecedentedly powerful AI computing power, which poses a challenge to the existing computer architecture. The second is AI infrastructure. Artificial intelligence products such as Wenxin Yiyan have complex computing scenarios and multi-dimensional computing architectures. Enterprises need a full-stack AI infrastructure that integrates chips, frameworks, models, and applications for AI development, and end-to-end intelligence. The whole process of closed-loop AI development. The last is the level of AI engineering. When AI generative products are put into industrial practice, they need to carry out secondary adaptation development based on specific business scenarios. However, the entire process of AI model development and operation is highly complex, and irregular operations in any link will affect the effect of the AI model. The level of AI engineering determines whether AI can break through the "last mile" of industrial implementation.
Facing the new rules of the game in the cloud market, only cloud vendors that solve the above three challenges for enterprises are eligible to get admission tickets. Shen Dou, executive vice president of Baidu Group and president of Baidu Smart Cloud Business Group, said that cloud services have transitioned from the digital age to the smart age. In the past, enterprises chose cloud vendors more based on basic cloud services such as computing power and storage. Demand will focus more on intelligent service levels.
Wenxin Yiyan will be released soon, and Baidu Smart Cloud will upgrade three supporting cloud services
In the era of intelligent computing, enterprises are facing many challenges such as the above-mentioned shortage of intelligent computing power, the complexity of the whole process of AI development, and the difficulty of engineering the AI industry. Baidu said that it will provide the industry with three major core competencies to help the industry's intelligent transformation on the fast track.
First, it provides enterprises with a huge amount of high-performance intelligent computing power to support the industrial implementation of generative AI products such as Wenxin Yiyan. Baidu has currently built smart computing centers in Yangquan, Shanxi, and Yancheng, Jiangsu. Among them, Baidu Yangquan Intelligent Computing Center is the largest single intelligent computing center in Asia, with a construction scale of 4 EFLOPS (40 billion floating-point operations per second) AI computing power, which can meet the ultra-large-scale AI computing needs of various industries. Song Chunxiao, head of strategy at Kunlun Core Technology, said: "Artificial intelligence chips are the core of computing power. The second generation of Kunlun Core has been widely used in the application of Baidu Wenxin large models, and provides AI computing power for the intelligent upgrade of all walks of life. support."
It is understood that Baidu Intelligent Computing Center continues to innovate the computing architecture, supports the computing tasks of different business scenarios of enterprises in the era of intelligent computing, and improves the business efficiency and innovation capabilities of enterprises. At present, Baidu Intelligent Computing Center has supported cutting-edge AI applications such as Wenxin pre-trained large models, biological computing, and autonomous driving. At the same time, based on the self-developed innovative technology, the PUE can be as low as 1.08, realizing efficient and energy-saving operation, thereby reducing customers' electricity bills and operation and maintenance costs.
(Baidu Yangquan Intelligent Computing Center real picture)
Second, provide enterprises with new AI infrastructure to reduce costs and increase efficiency in AI development. Baidu Smart Cloud launched the "Baidu AI Base" by encapsulating computing power, framework, models, and AI applications. From the high-end chip Kunlun core, to the deep learning framework of the flying paddle, to the Wenxin pre-trained large model, and then to the AI application, an end-to-end intelligent closed-loop is realized, thereby turning the high-threshold AI technology into a power supply for enterprises like hydropower On-demand access, greatly reducing enterprise development costs and improving efficiency. Song Fei, General Manager of Baidu Intelligent Cloud Cloud Computing Product Solutions and Operations Department, said: "Baidu's AI base can be continuously optimized based on actual business data, so that the resource utilization rate can be increased to 70%, and the enterprise development efficiency can be increased by 100%."
Finally, it provides enterprises with AI R&D and operation integration (MLOps) capabilities, and accelerates the rapid industrial landing of generative AI and other large-scale model products. Xin Zhou, director of Baidu AI Central Taiwan, said that the Baidu AI base will provide a series of AI R&D and operation and maintenance tools for enterprises. After accessing Wenxin Yiyan, enterprises can easily and cost-effectively complete the adaptation and secondary development of business scenarios, and use AI engineering capabilities to help the industry break through the "last mile" of AI implementation.
With the support of low-cost multi-architecture intelligent computing power, "AI large base" new AI infrastructure, and MLOps AI engineering capabilities, Wenxin Yiyan will accelerate the arrival of industrial intelligence and bring true AI inclusiveness.