AI "iPhone moment" has arrived, and industry chain companies are rushing ahead
Tech company leaders talk about AI’s “iPhone moment” opportunities
A chat tool may trigger an industrial revolution. At the beginning of this year, ChatGPT3.5 came out, and the global artificial intelligence field was booming. Even NVIDIA founder and CEO Huang Jensen couldn't help but lament: AI's "iPhone moment" has arrived.
"ChatGPT-like will appear in China within two years, and the applications will be more creative." "The era of intelligence will usher in the great development of server chips, and the next three years will be a new round of reshuffle." "The AI singularity is coming, from switches to High-end AI servers, cloud computing and data centers will all enter a new growth cycle” - Recently, at the “Dialogue with Leaders” boutique listed company exchange conference jointly hosted by GF Securities and Shanghai Securities News, representatives from VeriSilicon, The heads of technology companies such as Fii, Jewart, and Shenghong Technology talked about the opportunities of AI "iPhone Moment".
Under the boom of large models, many listed companies in the industry chain are running on the road to solidifying the foundation of the AI industry with their "hard power", laying a solid foundation for the subsequent implementation of large models and applications.
AI reaches the singularity moment
Entering early summer, ChatGPT is becoming more and more popular. The advent of ChatGPT can be called the "iPhone moment" of AI technology, and even Huang Renxun sighed.
Looking back at the development of smartphones, after the iPhone was born, it not only led the development of the smartphone industry, but also created a magnificent mobile Internet era.
"At present, AIGC (generative artificial intelligence) has completed its transformation from 0 to 1, and the next step from 1 to 10 will be the 'thousand model war'; within two years, a large number of ChatGPT-like applications will appear in China, and the core applications will be more creative. Dai Weimin, the founder, chairman and president of the original shares, made a bold prediction in his keynote speech at the exchange meeting. Dai Weimin believes that just like the iPhone led the development of smartphones and mobile Internet, ChatGPT will lead an AI industry revolution , the current ChatGPT4.0 is similar to college students, with the ability to learn and solve problems in all subjects. In the next step, large models in vertical fields will be like masters and doctoral students, and the ability to transform knowledge in professional fields will be more in-depth. By then, AI will transform from improving efficiency to Create value in niche industries.
However, Dai Weimin believes that unlike smartphones first and then the mobile Internet "bull market" starting, the AI revolution triggered by ChatGPT will first unfold in the fields of software and applications, and then extend to hardware upgrades. Corresponding to the capital market, the next round of bull market may be "soft first and then hard", with the market taking the lead in the software field.
"Generative large models are the future of AI." Suiyuan Technology, which focuses on artificial intelligence cloud computing chips, has a deep understanding of the AI wave triggered by ChatGPT. Jiang Yujing, vice president of Suiyuan Technology, said in his speech that large models put forward higher requirements for hardware technologies such as computing power, memory and bandwidth; the development path of AI will be from basic large models supported by large computing power to After a certain stage, it will be combined with the industry’s large scene models and application models and develop rapidly. In the future, general artificial intelligence will become a trend-setter in the market.
When it comes to AI’s demand for hardware, entrepreneurs such as Fii CEO Zheng Hongmeng and Shenghong Technology Chairman Chen Tao have deep feelings.
“Training a large model requires hundreds of millions of dollars.” Zheng Hongmeng said that data, algorithms/models, and computing power are the three key elements for the development of AI. Computing power is based on AI data centers, AI servers and accelerators, high-speed transmission and high efficiency. Storage is the core element for the optimization and evolution of large AI models; computing power chips are the core of the "electronic AIGC" industry.
Chen Tao said that AI applications are gradually being implemented in the fields of images, games, machine vision and other fields, and the massive data generated require higher computing power and storage, which is critical for high-end HDI (high-density interconnection circuit boards), high-speed and high-frequency PCB (printed circuit board) products have generated strong demand. Zhou Xunwei, chairman of Jiehuat, believes that in the era of intelligence, analog chips will achieve tremendous development, and the next three years will be a new round of reshuffle in the analog chip industry.
Talking about the implementation of AI applications, Yin Jun, vice president of Dahua Co., Ltd. and director of the Advanced Technology Research Institute, said that the integration of GPT technology and industry will greatly empower the digitalization and intelligent upgrading of many industries such as smart IoT; industries The large model is like deploying a dedicated brain in each industry. It will develop collaboratively with the industry like industry experts, and use the industry's digital capabilities to form new industrial productivity based on intelligent forms.
"In the field of training, the current level of most teachers may not be as high as ChatGPT.". "Editor CEO Han Xiangyang described the intelligence level of ChatGPT in this way. He is optimistic about ChatGPT's unlimited application prospects.
Companies in the industrial chain are “jumping away”
At the moment when AI reaches the singularity, the starting gun for the industry has not yet sounded, and many far-sighted semiconductor companies and AI application companies have jumped ahead.
Before the birth of ChatGPT, VeriSilicon had already applied the Transform algorithm to the visual field. "Dai Weimin gave an example. The company's customer Lanyang Intelligent used VeriSilicon's multiple processor IPs to deploy high-performance artificial intelligence chips based on the scalable Chiplet architecture, targeting data centers, high-performance computing, automobiles and other fields.
Dai Weimin also introduced that the demand for network communication speed and performance in data centers continues to increase, and high-speed interface technology is entering a critical development period. It is expected that SerDes (serializer/deserializer) related technologies will usher in rapid development in the future. As early as February 2021, VeriSilicon signed a cooperation framework agreement with Alphawave, a leading company in the high-speed SerDes field. It has the exclusive right to sell a series of Alphawave's multi-standard SerDes IP in China, and has become Alphawave's first choice globally. ASIC partners.
"Overseas, the company provides server-related PCB products to North American customers." Chen Tao introduced that supply concentration is low, downstream applications are broad, and technology categories are diverse... The PCB market, which exceeds 80 billion US dollars, is exploding due to AI. More opportunity. The main driving force for the company's performance growth will be in application fields such as new generation communication technology, artificial intelligence, digital economy, cloud computing and data centers. Currently, Shenghong Technology’s order visibility is approximately one month.
It is understood that Shenghong Technology has become an industry leader in the fields of GPU (graphics processor) independent display, small spacing LED, Mini direct display, smart speakers, broadband access, new energy vehicles and other segments, and has entered the AI super computing, cloud servers, base station AAU and other fields.
The rapid development of ChatGPT indicates that capital expenditures related to AI computing power may rise. The company expects to continue to drive sales of AI servers in the second half of the year and benefit from market expansion brought about by ChatGPT and others. Zheng Hongmeng said that due to the strong demand in the AI market, Fii Industrial has achieved improvements in various businesses and gross profit margins in the first quarter of this year. Among them, 5G high-speed network (5G/switch/WiFi) and AI computing power (AI server/HPC high-performance computing) business have achieved double-digit growth.
AI painting tool Midjourney, AI search... Artificial intelligence is rapidly being implemented in multiple industries, empowering thousands of industries, and constantly creating new work scenarios.
Talking about the impact of AI on the industry, Jiang Yujing gave an example that the proportion of data generated by generative AIGC will increase from 1% to 10% by 2025. AIGC is regarded as one of the five major technologies by Gartner, an international research institution. It is expected to achieve large-scale application within 2 to 5 years. The development of new drugs will usher in a new era, as machine learning technology accelerates the development of chemicals (substances). It is reported that for AI intelligent computing centers, Suiyuan Technology has launched a series of cloud AI computing chips and Yunsui computing accelerator cards such as Suisi, providing full-stack cloud training and reasoning products and solutions for multiple industries and complex scenarios.
Dahua Technology provides end-to-end machine vision solution delivery, covering the entire process from hardware to algorithms. "Yin Jun said that Huarui Technology, a subsidiary of the company, has been working in the fields of machine vision and mobile robots for many years, and has formed optical imaging, high-precision image processing, visual algorithm programming, embedded heterogeneous computing, and ultra-low power miniaturized hardware design. , cluster scheduling, autonomous navigation, intelligent battery management system and other core technologies.
The above is the detailed content of AI's 'iPhone moment” has arrived, and industry chain companies are rushing ahead. For more information, please follow other related articles on the PHP Chinese website!