With the rapid iteration of artificial intelligence technology and the increasing expansion of application scenarios, AI concept stocks in the secondary market have ushered in a new round of gains.
At the beginning of the year, Sora’s stunning performance ignited the enthusiasm for AI concept stocks. Following this, the AI application sector led by domestic long text and large model concept stocks rose strongly last month, with many stocks hitting their daily limit. Not long ago, the advent of the music platform Suno has inspired innovative thinking in music concept stocks. Entering 2024, with the in-depth advancement of large models in all walks of life, it is expected that more complex and diverse application scenarios will gradually be realized, frequently affecting the nerves of the secondary market.
A year and a half ago, the release of ChatGPT triggered widespread global attention to artificial intelligence and triggered a new round of AI arms race among global technology giants. The domestic market is not far behind, with companies competing fiercely around core algorithms, model parameters, computing capabilities and other aspects. After the technological explosion in the “deep learning framework optimization” and “large-scale pre-training model development” stages, the AI world is about to usher in huge changes and a new round of competition.
In 2024, OpenAI released the Wensheng video large model Sora, launching the full implementation of AI large models in the film and television industry. Not to be outdone, Google also launched Gemini and announced that it would implement a series of intelligent functions by applying a derivative version of Gemini Nano to smartphones. Subsequently, the SunoV3 model was born, starting a new round of revolutionary thinking in the music field. The industry predicts that 2024 will usher in the first year of the explosion of AI applications. AI will be implemented in all walks of life, and the wave of AI commercialization will accelerate.
Looking at the domestic market, Kunlun Wanwei was the first to speak out in this wave. On April 17, one year after the official release of the self-developed 100-billion-level large language model “Tiangong”, Kunlun Technology released the world’s largest open source MOE large model with 400 billion parameters – Tiangong 3.0. This model uses a 400 billion-level parameter MoE hybrid expert model, which is one of the open source MoE models with the largest model parameters and the strongest performance in the world. At the same time, Tiangong SkyMusic, China’s first large-scale AI music generation model based on Tiangong 3.0, was also released.