Artificial intelligence has entered a new stage of development and is gradually moving towards AGI; AIGC creates a new revolution in human-computer interaction and will usher in more new opportunities; The general large model is the key to the horizontal division of labor in the AI industry, which is expected to achieve scale effect and flywheel effect, and the business model is gradually clear. Generative AI has the potential to disrupt the existing competitive landscape for traffic.

Overseas market is the first to start, overseas deductive logic: computing power (the first to start, the largest increase) - large model - application, domestic deductive logic: data - computing power, large model - application.

Summary of investment logic:

【Phase 1】Starting point: 100 model wars, giants and vertical manufacturers have invested in the research and development and training of large models, early training has a large demand for high-performance computing power, manufacturers hoard a large number of chips, the certainty of computing power investment is relatively high, and large models and applications are still in the early stage.

[Phase 2] under pressure: path differentiation, a small number of general base large models + more vertical industry/company large model manufacturers, based on open source model development and underlying self-development in parallel, NLP large model training computing power demand is limited, the application side has not yet ushered in substantial growth, inference computing power consumption is limited, large models are also in the early stage of commercialization, and the overall industrial chain is under pressure.

[Phase 3] Breakthrough: Technology iteration, combined with new hardware to bring new changes, multimodal gradually mature, 3D generation, strategy generation and other technologies gradually breakthroughs, in the future, combined with MR, robots and other new hardware, the application will accelerate the popularization, further open up the commercialization space, and drive the overall industrial chain to form a positive cycle.

On the one hand, the iteration of large models and AI applications will continue to bring about changes in computing power demand and computing power structure, and on the other hand, the evolution of the industrial chain from basic platforms, to infrastructure, and then to AI applications will bring corresponding industrial opportunities.

It is recommended to pay attention to: 1) chip-related enterprises; 2) Enterprises related to the server industry chain; 3) Application-related enterprises.

Investment logic

The transformation of human-computer interaction has brought industrial investment opportunities

Progress of Artificial Intelligence Development

Classification of AI business models

The commercialization of large models has just begun, and small models have achieved a closed loop of commercial value in some fields

大模型可能打破流量竞争格局

人工智能的飞轮效应

相比于互联网的平台效应，AI有两个较显著的特点，一是规模效应，二是涌现能力。规模效应指的是AI模型的性能会随着模型规模增长而增长，包括参数量和数据量。涌现能力，指的是当模型足够大、数据足够多的时候，模型自己就会涌现出全新的能力。

人工智能投资逻辑

大模型厂商的飞轮效应

Artificial Intelligence Investment Map