在近日举办的2023百度云智大会·智算大会上，百度集团副总裁侯震宇发表了重要演讲，指出大模型正在重构云计算，推动AI原生云的发展。
据侯震宇介绍，进入AI原生时代，云计算基础设施体系需要进行全面重构，以满足大规模模型训练的需求，为构建繁荣的AI原生生态提供坚实的基础。他强调，大模型重构云计算主要表现在三个方面：一是AI原生云将改变云计算的格局；二是MaaS（模型即服务）将成为新的基础服务；三是AI原生应用将催生新的研发范式。
首先，侯震宇认为，大模型将重塑云计算的架构和功能。传统的云计算更多的是为了支撑应用程序的运行，而大模型的出现使得云计算更加强调对数据的处理和分析能力，从而更加符合AI技术的发展需求。
其次，MaaS作为新型的服务模式，将逐渐取代传统IaaS（基础设施即服务）、PaaS（平台即服务）等形式，成为云计算的新常态。通过MaaS，用户可以方便地使用预训练好的模型，而无需关心底层的硬件设施和软件系统。
最后，随着AI原生应用的不断涌现，新的研发范式也将随之诞生。这些应用不仅能够更好地利用大模型的能力，还能够在开发过程中实现更高的效率和质量。
总的来说，大模型正在推动云计算向AI原生云转型，这将为人工智能技术的发展带来巨大的机遇和挑战。在未来，我们有理由相信，云计算将在AI原生时代的引领下，迎来全新的发展篇章。
英语如下：
News Title: “Baidu’s Hou Zhunyu: Large Models Will Reconstruct Cloud Computing, AI-native Cloud Will Become a New Trend”
Keywords: large models, cloud computing, AI-native
News Content:
At the 2023 Baidu Cloud Intelligence Conference – Intelligent Computing Conference held recently, Vice President of Baidu Group, Hou Zhunyu delivered an important speech, pointing out that large models are reconstructing cloud computing and driving the development of AI-native clouds.
According to Hou Zhunyu, with the arrival of the AI-native era, the cloud infrastructure system needs to be completely reconstructed to meet the demands for massive model training, providing a solid foundation for building a thriving AI-native ecosystem. He emphasized that the reconstruction of cloud computing by large models mainly manifests in three aspects: First, AI-native cloud will change the landscape of cloud computing; secondly, MaaS (Model-as-a-Service) will become a new basic service; thirdly, AI-native applications will give rise to new research paradigms.
Firstly, Hou Zhunyu believes that large models will reshape the architecture and functions of cloud computing. Traditional cloud computing is more focused on supporting the operation of application programs, while the emergence of large models makes cloud computing place greater emphasis on data processing and analysis capabilities, thus better fitting the development needs of AI technology.
Secondly, as a new mode of service, MaaS will gradually replace traditional IaaS (Infrastructure-as-a-Service), PaaS (Platform-as-a-Service), etc., becoming the norm of cloud computing. Through MaaS, users can conveniently use pre-trained models without worrying about the underlying hardware facilities and software systems.
Finally, with the continuous emergence of AI-native applications, new research paradigms will also be born. These applications not only make better use of the capabilities of large models but also achieve higher efficiency and quality during the development process.
Overall, large models are driving cloud computing towards AI-native clouds, which will bring huge opportunities and challenges to the development of artificial intelligence technology. In the future, we have reason to believe that cloud computing will usher in a brand new chapter under the guidance of the AI-native era.
【来源】https://www.bjnews.com.cn/detail/1703048730169774.html
Views: 2