Today, CPO and multi-modal AI had the highest gains.

As of press release, Tengjing Technology, Tianfu Communications, and Bochuang Technology have risen by 20CM, Yuanjie Technology, Zhongji Xuchuang, and Saiwei Electronics, etc. have risen by more than 10%, Wufang Optoelectronics, Cambridge Technology, Huagong Technology, Voice Communications, Suzhou Keda, Shengshi Technology, Netda Software, Xinyisheng, Liante Technology, etc. have risen by more than 8%, and Yuncong Technology, Danghong Technology, Kunlun Wanwei, 360, etc. have followed suit.

According to the news, the search engine Google released the large multi-modal model Gemini, which sparked widespread debate, and the company's market capitalization surged by more than 80 billion US dollars overnight.

Is Google falsifying?

On Wednesday, Google unveiled the Gemini, a new generation of large models. As a large multi-modal model, Gemini can simultaneously recognize and understand five types of information: text, images, audio, video, and code, and the understanding of the information is very accurate.

Gemini includes three versions: Ultra, Pro, and Nano. Nano can run offline on the mobile side, Pro can achieve more advanced reasoning and planning (supports Bard), and Ultra, the most powerful, will open to the public early next year.

According to the demo video “Close Interaction with Gemini: Interacting with Multi-Modal AI” posted by Google on YouTube,Gemini gave very smooth and flexible feedback on prompts such as images and voices, and its power was amazing.

Affected by this, Google's stock price jumped high on Thursday and closed up 5.34% to $138.45 per share. Market capitalization surged by more than 80 billion US dollars in one day, while the valuation of its competitor OpenAI was probably between 80 billion and 90 billion US dollars.

At the same time, questions followed. Some media and users have discovered after using it for themselves that Gemini doesn't achieve the results shown in the video.

Google also acknowledged that the presentation was not live or voice-over.“For the purposes of this presentation, for the sake of brevity, the latency has been reduced and the Gemini output has been shortened”.

In an interview, Demis Hassabis, CEO of Google DeepMind, emphasized that Gemini is a new breed of artificial intelligence —“Native multi-modality”, which is different from the existing “patchwork multi-modal” model, that is,It was trained from the start using a variety of modes, such as audio, video, and images.As a result, Gemini has paved an unprecedented path in artificial intelligence and could bring about a major new breakthrough.

Institutions are optimistic about computing power and applications

Although Gemini may not be as powerful as in the demo video, investors are amazed by the rapid development and potential of the AI industry. At the same time, the investment opportunities in it cannot be ignored.

Looking forward to the future, Orient Securities believes that the focus of conquering in the next stage will definitely be multi-modal technology. The release of Google Gemini has shown the world a variety of application scenarios for the multi-modal model. The future still needs to be continuously explored, and the time for ChatGPT in the multi-modal field is yet to come.

CITIC Securities believes that the official release of Google's multi-modal Gemini modelOn the one hand, it can expand application scenarios, and on the other hand, it can bring about a continuous upgrade in computing power requirements. Continuing to be optimistic about the future of the AI industry, the subsequent release of models such as GPT-5 will also bring more catalysts.

Ping An Securities pointed out that competition among artificial intelligence giants in the field of large models is still heated. Against this background, they continue to be optimistic about the growth in AI computing power demand. At the same time, the current intense competition for multi-modal AI will also continue to raise the overall capability level of big models, helping to continuously expand the application scenarios and boundaries of big models, and leave a broad space for imagination.

On the target side: 1) In terms of computing power, I recommend Haiguang Information, Zhongke Shuguang, Ziguang Shares, Wave Information, and Longxin Zhongke; I recommend focusing on Shenzhou Digital, Cambrian, and Jing Jiawei; 2) In terms of algorithms, I recommend iFLYTEK; 3) in terms of application scenarios, I recommend Jinshan Office, and I recommend focusing on Wanxing Technology and Thors; 4) In terms of cybersecurity, I strongly recommend the star Qichen.

Open Source Securities believes that the AI multi-modal model represented by Pika 1.0 and Gemini continues to break through, orPromote the application of large-scale models in the fields of tools, education, office, e-commerce, marketing, etc., to accelerate implementation and open up commercialization space.It is recommended to continue to lay out AI applications:

1) The target benefiting from “AI+ tools” is Kunlun World Wide. 2) The target benefiting from “AI+ education” is Century Tianhong. 3) “AI+ Music” focuses on recommending Shengtian Network. 4) The target benefiting from “AI+ proofreading/animation” is Guomai culture. 5) The target benefiting from “AI+ Gaming” is Palm Fun Technology. 6) The beneficiaries of “AI+ Marketing/Virtual People” include Blue Cursor, Insai Group, Yuanlong Yatu, Gravity Media, etc. 7) The beneficiaries of “AI+IP/film/video” include Shanghai Film, Aofei Entertainment, Chinese Online, Huace Film and Television, Jebsen Co., Ltd., Wanxing Technology, and Easypoint World.