AIGC Industry Panorama Report 2023

Since Google proposed the Transformer architecture in 2017, AI development has gradually stepped into the era of pre-training large models.In June 2018, the number of OpenAI's GPT model parameter has reached 117 million, and the number of model parameter has begun to realize the development of a billion base flyover, which on average doubles every 3-4 months, thus bringing the demand for training arithmetic is also "At the end of 2022, after OpenAI's GPT model emergence capability, the AI industry rapidly entered the AIGC era, which is technically supported by large models.

AIGC Industry Market Size

AIGC (AI-Generated Content) refers to a new type of content production that utilizes artificial intelligence technology (generative AI path) to generate content.20 ChatGPT, an AIGC app that went live in November 22, has rapidly gained a large number of users by virtue of its excellence in the fields of semantic comprehension, text authoring, code writing, logical reasoning, and knowledge quizzing, as well as its natural language dialog with low threshold interaction, it quickly gained a large number of users and exceeded 100 million monthly activities in January 23, breaking the growth rate record of former consumer-grade applications. Microsoft says it sees the beginnings of AGI (Generalized Artificial Intelligence) in GPT-4 (the big model running behind ChatGPT Plus). New forms of AIGC applications such as Midjourney have appeared in the daily life and work of the public, and new possibilities have been seen for the intelligent upgrading of various industries, further expanding the imagination of "AI industry" and "industrial AI".

The technical support of AIGC application innovation is "Generative Adversarial Network (GAN)/Diffusion Model (Diffusion)" and "Transformer Pre-training Large Model" two types of large model branches. AIGC applications to show the energy of the big model at the same time, enterprises have also strengthened the layout of related products and technologies, cloud vendors, AI factories, startups, various industries and technical service providers and other players in various fields of the industry have released a big model or a big model based on the application of products and various types of technical services.

Reconfiguring the AI Development Deployment Paradigm

In the past, data centers were mainly leased and self-built, with the arithmetic demand side choosing to lease or self-build based on its own business volume, financial budget, data privacy requirements, and so on. In the context of the AIGC era, data centers will be configured with more AI servers to meet the increasing demand for intelligent computing power, cloud vendors are proposing MaaS (Model as a Service) model as a service business model, cloud computing, intelligent computing power, modeling capabilities, and other resources to do a high degree of integration, the customer can be called directly in the cloud, development and deployment of the model, a better fit for the customer's personalized needs. personalized needs.

AIGC will trigger deep changes across the industry

AIGC mainly affects content creation and human-computer interaction, so the higher the degree of linearization of the value chain, the higher the proportion of content in the value chain, the more obvious the subversive effect of AIGC on it; on the other hand, the industry's own characteristics such as data, knowledge, and regulatory requirements will also profoundly affect the penetration speed of AIGC technology. For example, industries such as e-commerce, games, advertising, film and media, etc. that take content production as the core of value, and industries such as e-commerce, finance, etc. where R&D, design, marketing and other links have a high position in the industry value chain, can quickly see the substitution of AIGC applications on the original production tools and the change of business processes.