Industry Trends | The implementation of application-side AI, looking at the future of cloud computing from North American giants

Every reporter: Liu Mingtao Every editor: Ye Feng

North American cloud computing giants are actively embracing this wave of AI. The integration of AI capabilities will also most likely bring medium and long-term growth momentum to cloud vendors’ infrastructure, model optimization, application basic software and other fields.

ChatGPT has been successful in the global market and triggered an "arms race" among the world's major technology companies in the field of AI. As the most ideal carrier at the computing level, the cloud platform is expected to continue to benefit from the development of the global AI industry. Currently, the three major cloud computing giants, Amazon, Google and Microsoft, have launched products related to LLM and generative AI. However, there are certain divergences between the three major cloud vendors in terms of layout ideas and commercialization paths. Microsoft and Google rely on closed-source models and strive to create a closed-loop AI ecosystem, while AWS has a relatively open mind and tries to conduct AI and AI from different angles. The integration of cloud computing.

Starting in March 2023, Microsoft and Google have added a large number of services based on AI model capabilities to their respective cloud products, and provide closed-loop AI services from hardware to models to upstream applications. In contrast, Amazon focuses on single-point changes in its product structure, providing products such as Amazon Bedrock (a convenient API service in the process of building large language models), and focusing more on open source models and related hardware development. middle. Fundamentally, what determines the current different approaches of Microsoft, Google, Amazon, and Meta to AI capabilities is the large difference in language model technical capabilities among the current manufacturers.

Microsoft and Google take advantage of their relative advantages in technical capabilities to transform open source into closed source, and strive to create a closed AI ecosystem with underlying model capabilities as the core. Amazon and others are more inclined to cultivate open source communities to jointly iterate models to narrow the gap, pay attention to underlying hardware research and development, and explore the possibility of integrating AI and cloud from multiple angles.

According to the computing products of the three major cloud service providers, the price of computing products using GPU acceleration is generally 3 to 6 times the price of CPU products. Considering that the subsequent introduction of H100 products is likely to increase the overall unit price. On the computing load side, according to the financial reports of various companies and comprehensive calculations by Trendforce, the load proportion of GPUs and other AI chips in the entire computing business line of the three major cloud vendors' businesses is less than 10%.

Based on the calculation of cloud vendors’ income and IT load structure, CITIC Securities predicts that every 1% increase in AI chip load will bring additional revenue of US$1.2 billion, US$700 million, and US$400 million to AWS, Azure, and GCP respectively. , if the penetration rate increases to 10% in the next 2-3 years, it is expected to bring nearly US$20 billion in additional revenue to the three major cloud vendors each year. And the growth in data volume is also expected to bring about growth in demand in storage, networking and other fields. In addition, cloud service providers will introduce AI products into different business areas, which will also reduce costs in advertising, operation and maintenance, etc.

The application of AI will long-term promote the development of cloud service providers in areas such as infrastructure, model optimization and basic software. In the short and medium term, it is expected that the most difficult time for North American cloud vendors' revenue growth is about to pass, and it is expected to gradually stabilize and recover in the second half of the year. As overseas technology giants and software companies continue to introduce and commercialize AI, cloud vendors are expected to continue to benefit.

Here, by integrating the latest research reports of more than 10 securities companies such as Tianfeng, Anxin, and Guosen, 4 companies are introduced for reference only.

1. Qingyun Technology

The company has broad cloud computing service capabilities and can meet the deployment needs of private cloud, public cloud and hybrid cloud according to customer needs. The products are being accelerated in finance, education, medical and other industries. In the future, the company is also expected to rely on its own service capabilities and technology First-mover advantage, we will continue to consolidate our dominant position, and we can expect future growth.

2.Kip Cloud

The company joined forces with Sichuan Province to create "digital people" to help open a new chapter in the Sichuan government affairs metaverse. The government metaverse is the company's core business that continues the past and the future. The company launches a new generation of digital finance platform and low-code development environment creation platform, which comprehensively covers the "service form", "service experience", "service entrance", "service location", Five major areas of "Service Portal". At the same time, with the requirements of carbon peaking, carbon neutrality strategy and digital transformation, big data, AI, etc. have a broad market space in the energy industry. The company is actively expanding its layout in the field of digital intelligence sources and expanding through extension through the acquisition of Tianyi Dataju. Improve the Kaipu Cloud technology system and enrich the strategic development of digital intelligence.

3, HKUST Guochuang

Since the company has long provided digital transformation services to large customers such as operators and energy groups, and has a broad base of potential customers, we can look forward to the expansion of its energy storage business. In terms of cloud platform, the company uses the "platform service" model to innovate digital logistics application scenarios, providing digital ETC and digital supply chain services to cargo owners, logistics companies and truck drivers, cultivating new growth points for the company.

4. Zhongke Information

According to OpenAI, data is one of the driving forces for AI model optimization. The company adopts a "general model-specific scenario" model, using the CV model as the main line technology to derive subdivided scenarios such as smart government affairs, smart manufacturing, and smart cities. , the company's main customers cover central and state agencies, provincial and municipal government agencies, and have accumulated a large amount of high-quality data for subsequent model iterations.

Daily Economic News

The above is the detailed content of Industry Trends | The implementation of application-side AI, looking at the future of cloud computing from North American giants. For more information, please follow other related articles on the PHP Chinese website!