If 2023 is considered the inaugural year for generative AI, then 2024 marks a year of accelerated popularization for this technology.

Following the release of ChatGPT, artificial intelligence has entered the era of large models. The emergence of generative AI is propelling humanity into the stage of digital civilization.

AI is drawing enterprises into a wave of transformation, as stated by Guo Wei, Chairman of Digital China: "In the next decade, the corporate strategy of all enterprises will fully leverage the three natives (cloud-native, digital-native, AI-native) to disrupt their own businesses. Digital China is also willing to become a partner throughout the entire lifecycle of corporate digital transformation, let us welcome this great era together."

Advertisement

01

Innovation in Information Technology and AI Computing Power, Laying the Foundation for Intelligent Computing

Innovation in information technology, also known as the information technology application innovation industry, is the foundation of data and network security, an essential component of new infrastructure, and the "engine" for the digital upgrade across various industries. Based on this innovation, Digital China continues to build its own brand "KunTai", forming three major product lines covering "data computing, terminal products, data networks", and continuously deploying in "indigenous computing power + intelligent computing".

In the past year, a series of self-branded products such as the KunTai QunTai AI learning machine and the KunTaiA924 training and inference integrated AI server have been successively launched; 14 KunTai servers have passed the "General Server Government Procurement Demand Standard Special Topic Evaluation".

With the development of AI, the amount of computing power data and the scale of parameters used by large models are growing "exponentially", leading to an explosive increase in demand for intelligent computing power. IDC estimates that by 2026, the scale of intelligent computing power in China is expected to reach the level of 10^18 floating-point calculations per second, with a compound annual growth rate of intelligent computing power expected to reach 52.3% from 2021 to 2026.Digital China's Vice President and Chairman of Digital China Information Innovation Holdings, Han Zhimin, also stated: "The current development of AI faces significant computational challenges. In addition to internet companies having large models and powerful computational clusters, various industries also require a substantial foundation of computational power." A series of issues and challenges at the infrastructure level have also emerged.

On one hand, there are complex compatibility and utilization rate issues between and within intelligent computing clusters. Currently, domestic companies are in a complex business environment, and heterogeneous intelligent computing infrastructure has become inevitable. However, there is still a lot of room for improvement in the model floating-point utilization rate that heterogeneous intelligent computing clusters can achieve. For instance, when OpenAI was training ChatGPT4, the utilization rate of MFU was only between 32% and 36%. The current average industry utilization rate of MFU is between 30% and 40%, and it is extremely difficult to increase it to 50%. Consequently, Digital China KunTai has introduced the HISO heterogeneous intelligent computing scheduling operation platform and the brand-new heterogeneous intelligent computing fusion acceleration platform HICA.

The Heterogeneous Intelligent Computing Scheduling Operation Platform (HISO), based on 100% cloud-native technology, integrates GPU hard partitioning and virtual partitioning technology, achieving the pooling of GPU resources and cross-cluster scheduling capabilities, thereby enhancing the resource utilization rate of GPU server clusters.

The Heterogeneous Intelligent Computing Fusion Acceleration Platform (HICA) features multi-core support in one cloud, compatible with mainstream AI chips from Huawei Ascend, NVIDIA, Intel, and other domestic and international brands. It can achieve mixed training and inference tasks on intelligent computing clusters composed of different brands and models of chips, with an expected reduction of 20% in idle GPU computational power.

On the other hand, generative AI models have a large number of parameters, and computational power and high energy consumption have become core issues that need to be addressed. To meet the computational demands of intelligent computing centers, Digital China KunTai has created a variety of AI servers, covering all scenarios from training to inference.

In response to the high energy consumption pain point of intelligent computing centers, Digital China KunTai has also introduced an integrated delivery of all-liquid-cooled full cabinet, which requires no on-site installation and debugging, increasing efficiency by tenfold; the maximum single cabinet power is over 60kW, with one cabinet replacing the traditional 4 to 8 cabinets, and the energy efficiency ratio is 1.5 times that of domestic competitors.

In the past year, the construction of intelligent computing centers has been in full swing across China.

At the policy level, "new quality productive forces" were written into the government work report for the first time, the "AI+" initiative was fully launched, and there was a special meeting on "AI Empowerment for Industrial Renewal" convened by the State-owned Assets Supervision and Administration Commission, proposing: to accelerate the construction of a batch of intelligent computing centers.

In 2023, there are more than 100 new intelligent computing center projects nationwide, with budgets basically in the hundreds of millions. Since the national launch of the "East Data West Computing" project, traditional IT companies, cloud vendors, and telecommunications operators have been actively planning the layout of intelligent computing centers.Digital China deeply participates in the smart computing centers, with Digital China KunTai being one of the important participants in the "Kunpeng + Ascend" computing industry ecosystem, providing professional infrastructure hardware for smart computing centers. Currently, Digital China and Digital China Holdings have participated in the construction and operation of smart computing centers such as Changchun New District Smart Computing Center, Shenyang Artificial Intelligence Computing Center, Xiamen Kunpeng Supercomputing Center, and the Hong Kong SAR Government's large model smart computing center.

It is worth mentioning that in January of this year, in order to provide the most basic technical environment for ecological partners and focus on AI computing power, the Digital China Shenzhen Artificial Intelligence Computing Center project has officially started. It is expected that soon, this smart computing center equipped with the Digital China Wenshu platform will officially provide services related to computing power, models, applications, and solutions to the market and the ecosystem.

Looking at the financial reports, the entry of AI and the synergy of products have led to explosive growth in Digital China's information innovation business. In 2023, the overall revenue of Digital China's information innovation business exceeded 3.4 billion, maintaining high growth for three consecutive years. In the first quarter of 2024, the revenue scale of Digital China KunTai's AI computing power-related business for a single quarter approached the annual total of the previous year.

02

AI-native scene empowerment, "To the vast and to the meticulous"

Over the past year, AI large model technology has developed rapidly. How to use large models to reduce costs and increase efficiency, and promote business growth, has become a real concern for enterprises.

In this era of the "hundred model war" or even the "thousand model war," how to transform technology into real productivity? The answer is to break the deadlock with the power of the ecosystem and accelerate the landing of generative AI at the enterprise level.

In 2023, the launch of Digital China Wenshu became a new paradigm to help enterprises land AI.

Currently, Digital China Wenshu has docked with dozens of mainstream large models and released agile applications such as Smart Reading, Smart Parsing, and Smart Q&A, successfully helping customers in the pharmaceutical and retail industries to land generative AI application scenarios.

At the 2024 Shuyun Yuanli Conference, a new version of Digital China Wenshu was released, including three major modules: Agent Engineering, Enterprise Knowledge Governance, and Model Training and Management.Taking the three major functional modules of the new version of Shenzhou Wx as the entry point, Shenzhou Digital has its own considerations: In reality, enterprises have a multitude of scenarios that require AI empowerment. For businesses, how to better integrate AI scenarios with each business, and how to enable internal IT and business departments to cooperate more intelligently, cannot be met by a simple developer platform alone. Therefore, enterprises need a proven efficient and comprehensive AI application development framework and specifications; they need the continuous accumulation of tools; the continuous deepening of knowledge governance, and even more so, the continuous training and optimization of several enterprise models. At the same time, to ensure data security, all of this needs to be completed within the private domain of the enterprise. In this regard, the AI native innovation of enterprises requires a platform to continuously empower and accelerate. Shenzhou Wx is such an AI native application empowerment platform, covering all customer needs of the enterprise.

Shenzhou Digital's Vice President and CTO, Li Gang, stated: "When 'extending broadly' has been fully expressed, Shenzhou Digital is even more determined to focus on the customer-centric 'pursuing the subtle', which is the real priority for AI to land."

Regarding the enterprise data security that everyone cares about, Li Gang shared in an interview that Shenzhou Wx protects enterprise data security through three aspects.

Firstly, the scenarios deployed by Shenzhou Wx are private. They are fully deployed in an environment that the customer can control. Secondly, the large models managed by Shenzhou Wx, whether commercial models or open-source models, are also locally deployed after fine-tuning and pre-training. On these two points, it can ensure the basic security issues. Thirdly, it is about how to ensure the accessibility of knowledge within an enterprise. Enterprises may need multiple models due to different departments, such as HR models, financial models, etc. The key is to give enterprise data a "security fence." Li Gang said: "Ensure that the model answers the correct questions it can answer, and does not answer irrelevant questions." By adding this point, it will further ensure the enterprise data security issues.

Additionally, we also see Shenzhou Digital fully leveraging its independent innovation capabilities in "computing power + AI" software and hardware products, creating the Shenzhou KunTai Wx all-in-one machine. Based on the KunTaiA924 training and inference integrated AI server, it can seamlessly connect training and inference, and carry out simultaneously, greatly improving the efficiency of large model training. Pre-installed with Shenzhou Digital's self-developed generative AI product "Shenzhou Wx", it is ready to use out of the box, with minimal deployment and no need for fine-tuning, allowing for rapid deployment of enterprise-specific AI large models.

03

AI industry acceleration, aggregation of ecological innovation

The emergence of Chat GPT has brought generative AI to everyone's attention overnight.For this revolutionary technology, every enterprise is in a state of anxiety, hoping to find opportunities for innovation within it. However, in the process of exploring AI, there are still many practical issues at hand.

On the demand side, many enterprises are at a loss when it comes to the iteration speed of generative artificial intelligence, and they need to find a balance between investment and return; on the supply side, in the process of commercialization, it is sometimes difficult for a single enterprise to combine the entire AI technology stack with the industry needs of customers; in terms of talent, the shortage of talent is the biggest obstacle to the continuous development of artificial intelligence.

Wang Bingfeng, Co-Chairman and CEO of Digital China, said: "In the face of these practical issues and the huge disruptive potential of artificial intelligence that we see in the future, no enterprise can go it alone. We believe that the ecosystem will become an indispensable and powerful driving force in the acceleration of AI implementation."

Digital China hopes to create a new type of ecological partnership, starting from three aspects: one is the ecosystem of solutions and technology, the second is the ecosystem of talent, and the third is the ecosystem of the market.

In terms of solutions and technology ecosystem, technology itself will not disrupt anything; only implementable solutions, applications, and business models will. AI-native applications can solve problems that could not be solved in the past. Only by working together to produce a sufficient number of AI-native solutions and applications can a healthy ecological environment be formed, allowing artificial intelligence to evolve from a corporate tool to an independent intelligent agent.

In terms of talent ecosystem, Digital China will join forces with the world's leading artificial intelligence enterprises to jointly build a training and certification system for artificial intelligence. At present, Digital China is the only one in mainland China that has official authorization from Microsoft, AWS, and Google as training partners, and is also a training partner for dozens of advanced enterprises such as Alibaba Cloud, Tencent, China Telecom, Huawei, Huawei Cloud, Oracle, and IBM.

In terms of market ecosystem, relying on Digital China's 2035 Lab and the Artificial Intelligence Research Institute, it continuously strengthens cooperation between Digital China and universities, research institutes, institutions, and associations to jointly promote long-term efforts in the technological ecosystem of artificial intelligence.

04

ConclusionAfter nearly a year of rapid development, generative AI continues to be a hot topic. The era of Artificial General Intelligence (AGI) has sounded its clarion call, and Digital China will focus on three things: empowering AI-native scenarios, multi-cloud heterogeneous green intelligent computing, and international AI ecosystem innovation.

At the end of the opening ceremony of the Digital Cloud Original Force Conference, Guo Wei, Chairman of Digital China (in his digital avatar), said: "We have proposed Digital China's new proposition for the AGI era: an AI landing acceleration engine centered on customers. The integration of data and cloud has also reached a new level, upgraded to AI-driven data-cloud integration. Digital China will continue to explore on the Chinese modernization path of 'Artificial Intelligence Plus,' using the International Innovation Center, a future landmark of AI, as a vehicle for more interaction and collaborative innovation."