.jpg)
Target group needs:
Front-end computing power requirements:
In Hong Kong, business scenarios involving large-scale use of artificial intelligence computing power are still dominated by large-scale model training, and the main customer groups for large-scale model training in Hong Kong are still mainly scientific research institutes. The main customer groups include the following three categories:
Research institutes:
For example, the Hong Kong Innovation Institute of the Chinese Academy of Sciences, Huawei Noah Laboratory, and the Hong Kong Institute of Applied Science. Through the development and cooperation of scientific research projects, we will deepen the scientific research cooperation between Guangdong, Hong Kong, and Macao, empower the development of a smart society in Hong Kong, and promote the development and inclusiveness of artificial intelligence capabilities.
Local Universities and Colleges in Hong Kong:
All eight universities in Hong Kong have AI research, and three of them are among the top 30 AI universities in the world. By leveraging the strong scientific research and technology strength and academic driving capabilities of local universities, we will develop and conduct industry research on artificial intelligence and large models, develop machine learning research and development, and help transform the results of artificial intelligence projects into applications.
Hong Kong's local financial and medical industries:
Realize artificial intelligence to empower local financial and medical industry applications. For example, apply artificial intelligence to develop medical computer vision image recognition and error detection related models, and help the financial industry realize financial forecasting and business intelligence analysis models, voice training models, etc.
Industry Overview:
Challenges facing Hong Kong’s AI industry
Insufficient computing power:
Enterprises face the challenge of "insufficient computing power" 44%
Large enterprises: 34%
Small and medium-sized enterprises: 47%
Current solutions to computing power
71% use cloud computing services
31% use mainland supercomputing centers
26% use overseas supercomputing centers
AI computing power development performance
AI computing power growth CAGR 28%
AI industry performance: ranked second for two consecutive years from 2021 to 2022;
Biotech financing scale: first in Asia and second in the world
Citations of AI papers: 3rd in the world;
Global Artificial Intelligence Most Innovative Cities Index (Source: AMiner)
Hong Kong Ranking :
Hong Kong ranks fifth in the world in the Scholar Index
Hong Kong ranks 5th in the world in terms of paper index
Hong Kong ranks 5th in the world in international index
Hong Kong ranks 11th in the world in innovation index
Institutional Index: Hong Kong ranks 20th in the world
Current status of Hong Kong’s AI industry (Source: 2023 Hong Kong Special Administrative Region Government Policy Address)
Three Hong Kong universities are among the top 30 AI universities in the world
The 2023-24 Budget announced a HK$3 billion allocation to strengthen basic research in the field of frontier technologies
AIR@InnoHK already has 14 AI research laboratories (manufacturing, healthcare, logistics, construction, etc.)
The 2022 Policy Address announced that HK$10 billion would be reserved to launch the Industry-Academic-Research 1+ Program
Since 2018, the government has invested more than HK$15 billion to support innovation and development
Leveraging a HK$30 billion co-investment fund and a HK$5 billion strategic innovation and technology fund to attract high-potential technology companies from the Mainland and overseas to set up operations in Hong Kong
Hong Kong’s regional advantages in AI development
Backed by the motherland
New DC system, diversified computing power supply, all-optical computing power network
Connecting the World
Strong, accessible, flexible, safe, agile
Central and Western Bridge
A bridgehead connecting the Guangdong-Hong Kong-Macao Greater Bay Area and the world
Hong Kong AI Education Development Strategy
The University of Hong Kong
There is a clear demand for computing power. Currently, the computing power scale that has been ordered exceeds 200P. We are negotiating with some hardware manufacturers on supercomputing cooperation such as computing power server provision .
The Hong Kong University of Science and Technology
We are currently planning to expand the computing power to over 500P, build a multi-campus computing power cluster network with the school-level platform as the center, and build efficient coordination capabilities between the Guangzhou campus and the Hong Kong campus. We are willing to carry out the adaptation of domestic computing power.
China Unicom and the Hong Kong University of Science and Technology have strategically cooperated to establish a joint innovation laboratory to develop and conduct industry research on artificial intelligence and large models, help transform the results of artificial intelligence projects into applications, jointly incubate high-value results in computing power scheduling, cross-border data, and AI empowerment, and cultivate high-level talents.
The Hong Kong Polytechnic University
The Hong Kong Polytechnic University has developed a key research project in 2023/24, "Cloud-Edge-Device Collaborative Computing Scheduling for AI Applications", which has received a grant of HK$3.6 million from the National Natural Science Foundation of China and the Research Grants Council. The project aims to develop a computing power management and computing-network convergence scheduling framework for cloud-edge-device collaborative computing networks to support resource-intensive and latency-sensitive artificial intelligence (AI) applications, and contribute to the construction of the national computing power network. The initial project period is 4 years.
The school-level scientific research platform has great potential for independent and controllable artificial intelligence computing power
One platform, multiple computing power
Build a school-level platform to flexibly meet the differentiated needs of each college
Full stack infrastructure software and hardware localization
Together we will accelerate the domestic production of the energy industry
Full stack construction
E2E solution planning, design and delivery, flexible matching of existing equipment room conditions, rapid deployment
Unified operation
Supports unified management and scheduling across multiple computing powers and heterogeneous clusters in different locations
Online/offline collaboration
Flexible access to computing power from third-party public computing power platforms, resources can be called on demand
Easy to use
Build an “out-of-the-box” app experience
Hong Kong AI Government Development Strategy
The Hong Kong government has once again emphasized in the 2023 Policy Address the acceleration of the construction of supercomputing centers. It is currently conducting a feasibility study on the establishment of an artificial intelligence supercomputing center, and has completed the consultation part of the bidding. The Hong Kong government has reserved 324 Nvidia H800 servers through Cyberport, with a total computing power of 3000P.
It is estimated that AI computing services will be provided in stages in 2025.
The new policy address proposes to further promote the "Hong Kong Innovation and Technology Development Blueprint", mentioning that Cyberport will set up an artificial intelligence supercomputing center in phases starting next year to support the computing power needs of scientific research and related industries and help the development of the artificial intelligence industry.
Scientific research advantages and industrial advantages: Scientific research innovation cannot be separated from the support of computing power infrastructure
High scientific research orientation
Global Innovation Center
Strong scientific research capabilities
16 national key laboratories
Abundant scientific and educational talents
Eight major universities, many universities and AI talents
Diverse application scenarios
Advantageous industries : finance, life sciences, microelectronics
Rich data accumulation
Digital society/digital government construction leads the world in data accumulation
AI enterprise resources are strong
It has already gathered and incubated a number of key artificial intelligence companies
Comments