The Zhitong Finance App learned that Cinda Securities released a research report saying that the trend of AI development is quite definitive. Currently, from a global perspective, the computing power industry chain is now more and more clear in the closed loop of upstream hardware, midstream server/switch, and downstream applications, showing a vibrant scene from the cloud side to the end side, and from hardware to software. It is recommended to focus first on the core manufacturers in the global computing power industry chain, followed by individual stocks related to end-side AI, which have a lot of potential. Furthermore, domestic alternative computing power stocks also have good opportunities. Relevant stocks include: [Cloud AI] Industrial Wealth Connect (601138.SH), etc.; [AI PC] Chunqiu Electronics (603890.SH), etc.; [AI Phone] Changxin Technology (300088.SZ), etc.

▍ The main views of Cinda Securities are as follows:

OpenAI releases Sora, and the film industry may face a time of dramatic change.

Open AI released Sora, which can generate videos up to a minute long. OpenAI is teaching AI to understand and simulate motion in reality to help humans solve the problem of interacting with reality. Sora is a generic visual data model that can generate videos and images of varying durations, aspect ratios, and resolutions, and can produce up to a full minute of high-definition video.

In terms of performance, first, Sora has selectable video length, resolution, and picture ratio. Traditional video and image generation usually cuts or trims video material to a standard size, while Sora trains the data at its native size. This approach brings flexibility and optimized framing and composition. Second, Sora has strong language comprehension skills. OpenAI first trained a highly descriptive subtitle model and then used it to generate text captions for training-focused videos. OpenAI also leverages GPT to turn short user prompts into longer detailed captions and send them to the video model, which allows Sora to accurately follow user prompts to produce high-quality videos.

Finally, in addition to using text, Sora can also use images and video as cues to perform various image and video editing tasks — creating perfectly looping videos, animating still images, scaling videos forward or backward in time, etc.

Google releases Genimi 1.5, which surpasses GPT-4 Turbo.

Genimi 1.5 is Google's largest model so far, and the Genimi 1.5 Pro version has now been released. Genimi 1.5 Pro comes with a contextual window of 128,000 tokens, and developers and enterprise customers can try out a contextual window of up to 1 million tokens in the personal preview version through AI Studio and Vertex AI. This means Genimi 1.5 Pro can process 1 hour of video, 11 hours of audio, over 30,000 lines of code, and 700,000 text at once.

In terms of architecture, Genimi 1.5 is based on research on Transformer and MoE architectures. According to paper data published by Google, in a horizontal comparison, Genimi 1.5 Pro's text, audio and video processing capabilities almost completely surpass GPT-4 Turbo. In vertical comparison, Genimi 1.5 Pro's performance greatly surpassed Genimi 1.0 Pro and slightly ahead of Genimi 1.0 Ultra.

Altman's trillion-dollar financing revealed, and the AI chip industry may be reshaped.

Ultraman is considering a project aimed at increasing global chip manufacturing capacity and has begun engaging with various investors, including the UAE government, according to the Wall Street Journal report on Thursday evening. If the financing is secured, it may reshape the semiconductor chip industry.

According to the Semiconductor Industry Association (SIA), the global chip industry's sales will grow 13% to reach 595.3 billion US dollars by 2024. This marks a significant increase compared to last year's 8% decline. In terms of market capitalization, if Ultraman can raise 7 trillion US dollars, he can even buy the entire chip industry, including Nvidia, TSMC, Broadcom, ASML, Samsung, AMD, Intel, etc.

Risk Factors:

Risk of macroeconomic fluctuations; risk of electronics industry development falling short of expectations; risk of exchange rate fluctuations.