news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

If you want to see the scenery of the Southern Song Dynasty, the famous painting "Four Scenery Landscapes" created by "the Saint of Painting" Liu Songnian is an excellent work. Today, the wind of digitalization is blowing in China's mountains and rivers. If you want to see the "national needs" and "Beijing needs" in the AI ​​era, the "AI Landscape" painted by Beijing Film Academy is worth seeing.

One year after its establishment, Beijing Electronics Digital Intelligence Technology Co., Ltd. (hereinafter referred to as "Beijing Electronics Digital Intelligence"), a subsidiary of Beijing Electronics Control, a super-large high-tech enterprise in Beijing, has launched four major AI innovative products:"Spark Intelligent Computing", "Forward AI Heterogeneous Computing Platform", "Pagoda Model Adaptation Platform", and "Red Lake Trusted Data Space".

This "Four Scenes of Landscape" in the AI ​​era contains the computing power, services, models, and data required for the development of the times. It is also one of the few "Chinese solutions" with full-stack AI tool capabilities. It can be called an AI panorama.

In recent years, we have seen that many manufacturers are working hard to expand the AI ​​ecosystem, continuously incorporating more AI technologies and products to expand their business territory, and moving from single roles such as hardware supply and cloud services to "full-stack AI service providers."

The reason is that AI service providers must transform themselves into a diversified and integrated ecological center, meet the complex and personalized needs of a large number of customers through "full-stack capabilities" and ensure the smooth progress of digital transformation.

What’s special about Nortel Digital Intelligence is that as an AI technology company that has been established for less than a year, its full-stack service capabilities are native to the AI ​​era.

Therefore, unlike ICT or cloud vendors that expand their ecosystem by expanding from their own advantageous businesses to other fields, Nortel Digital Intelligence has focused on empowering the industry in a systematic and full-link manner since its inception, which is equivalent to drawing a picture on a blank piece of paper that best suits the industrial ecosystem needs in the AI ​​era.

How to empower the industry with full-stack AI capabilities? Nortel Digital Intelligence, a native company in the AI ​​era, approaches the issue from a different perspective.

As a technology supplier, we must have a deep understanding of the needs of the times in order to solve problems accurately.

Why do governments and enterprises need "full-stack AI" empowerment to inject AI momentum into their businesses and organizations? Behind this is the changing and unchanging needs of the times.

Simply put, the users haven’t changed, but the calculations have.

With the advent of the AI ​​era and Moore's Law approaching its limit, computing tasks are moving from simple to highly complex.

Previously, computing was mainly general computing, with relatively simple computing power requirements and relatively simple infrastructure, mainly relying on data centers and CPU servers. The emergence of large-scale pre-trained models requires the use of a variety of different types of computing resources (such as CPU, GPU, FPGA, ASIC, etc.) to complete complex computing tasks.

At the same time, complex systems such as Wanka clusters and multi-heterogeneous clusters bring technical difficulties and require the intelligent computing center to provide systematic support. In addition to computing resources, they also require support from cluster management software, development tools, model libraries, etc.

It can be seen that the complexity of computing in the AI ​​era also requires governments and enterprises to have more dimensional and more difficult digital capabilities in order to keep up with the wave of intelligence in a timely manner.

At the same time, we found that the main players in the AI ​​era have not changed, and large government and Internet companies are still the pioneers of digital transformation. In particular, the government and central enterprises bear the responsibility of "leading the way" in developing the digital economy.

However, most governments and central enterprises are non-digital native organizations, and there is a certain gap between them and digital native Internet companies in terms of digital infrastructure, technical strength, digital talents, etc. Therefore, facing the complex computing needs of the AI ​​era, governments and central enterprises are more likely to encounter capacity shortcomings. Relying solely on purchasing computing resources, servers or model algorithms to supplement capabilities is not only inefficient, but also has a high total cost of ownership (TCO).Introducing AI in an integrated manner and obtaining ready-to-use solutions has become the choice of large government and enterprises in the AI ​​era.

Computing is changing, but the role of governments and state-owned enterprises as pioneers of digital intelligence remains unchanged. Therefore, the purchasing logic and purchasing decisions of AI must change, which will lead to changes in business models and make full-stack services a necessity.

Ultimately, AI needs to be delivered to digital participants such as governments and state-owned enterprises in a systematic and solution-oriented manner. This requires AI service providers to gather elements of the industrial chain and provide comprehensive empowerment.

So, how did Nortel Digital Intelligence, which was established a year ago, plan and implement its plans to become a full-stack service provider in the AI ​​era?

As a technology innovation company native to the AI ​​era, the greatest feature of Nortel Digital Intelligence is that it has insight into the needs of AI implementation in the industry from the very beginning, and has brought together AI industry elements with an open and diverse ecological concept.

Yang Zhen, head of Nortel Digital Strategy and Market, said that the biggest difference in the artificial intelligence industry is that no single company can accomplish everything independently, and it is necessary to use the ecosystem to aggregate everyone's strength to work together.

Therefore, Nortel Digital Intelligence created "Spark Intelligent Computing" based on the concept of "AI Factory".

Specifically,Spark Intelligent Computing can provide full-stack AI capabilities and fully empower the industry through performance-based billing, comprehensive large-model services, and special industry model services.In addition, local AI companies and talents are gathered through open AI exhibition spaces, roadshow spaces and joint laboratories.

Taking the Beijing Digital Economy Computing Center, a benchmark project created by Nortel Digital Intelligence, as an example, after its completion, it will not only provide highly cost-effective computing services to Beijing’s municipal and enterprise users, but also provide a platform for AI ecosystem partners to showcase scientific and technological achievements and exchange ideas.

It can be said that "Spark Smart Computing" reflects the thinking of Nortel Digital Intelligence as an AI native enterprise:By adopting a model of gathering production factors, cultivating production capabilities and fully empowering the industry, we serve AI industry ecosystem enterprises, further realize AI industry ecosystem operations, and make every Spark Intelligent Computing project a fertile ground and platform for the local artificial intelligence industry.

In view of the "three elements of AI" - computing power, algorithms, and data, Nortel Digital Intelligence has made a comprehensive product layout:

In terms of computing power, the "Forward AI Heterogeneous Computing Platform" will be used to promote the development of domestic computing power.

At a time when computing power is changing from "single computing" to "diversified and intelligent", the "Forward" platform, with its core technology of mixed heterogeneity, can achieve unified scheduling and management of domestic chips, stimulate the huge potential of different domestic computing power in various application scenarios, and achieve more efficient and cost-effective computing power supply. This means that ensuring that domestic chips can provide computing power for intelligent computing centers in an orderly and efficient manner is expected to reshape the competitiveness of domestic chips.

In terms of algorithms, in order to meet the needs of large models in various industries, Nortel Digital Intelligence has created the AI ​​tool chain "Pagoda Model Adaptation Platform", adapt the hardware downward to decouple and adapt the chip and large model extensively, so that the large model can run "easily and smoothly" on different AI chips; adapt the software upward to be compatible with major mainstream development frameworks, and lower the threshold for enterprises and developers to use it.

At the same time, in industries related to national economy and people's livelihood, such as government affairs, medical care, culture, education, and manufacturing, Nortel Digital Intelligence has laid out a vertical model matrix to further lower the threshold for model use and help industry AI innovation.

In terms of data, we have built a full-stack trusted data circulation infrastructure and data security facility, the "Red Lake AI Trusted Data Space", by making data trustworthy, controllable, available, and auditable, we create a safe and reliable operating environment, address the pain points of "inability to supply, inability to circulate, poor use, and high risk" in data value mining, and define the "AI endgame" with data trustworthiness.

It can be seen that Nortel Digital Intelligence has drawn a panoramic picture of AI with its full-stack capabilities covering big computing power, big models and big data, and a complete layout of the "three elements of AI".

Why did Nortel Digital Intelligence complete the concept innovation and capability building of a full-stack service provider in the AI ​​era in just one year?

We further learned that Nortel Digital Intelligence has its own full-stack AI methodology.

Everyone knows that technological innovation is important, which is why there are phenomena such as "innovation for the sake of innovation" and "shooting first and then drawing the target", which not only wastes R&D resources, but also consumes the public's enthusiasm for domestic technology. The way that Nortel Digital Intelligence does AI is "targeted":

1. Find the pain points first, then create solutions.

While all walks of life are calling for AI to advance rapidly, we cannot ignore that AI is not free. Introducing AI means that governments and enterprises have to increase costs and invest resources. Whether it can bring benefits within a reliable cycle is the most difficult question that technology companies must answer today. Especially for some vertical industries, which have their own uniqueness, the value of general AI is often limited, and the actual penetration rate of AI is not high.

In fact, the solution is very simple, which is to "find the right remedy before prescribing it." Taking Nortel Digital Intelligence as an example, many projects will first go deep into the industry and scenarios to find the pain points and difficulties in the customer's business, and then look for solutions that can truly reduce costs and increase efficiency.

For example, many hospitals are short of doctors and staff, but they don’t know what kind of AI products can solve this problem. When cooperating with hospitals, Nortel Digital Intelligence will deeply understand the actual needs of the scene, provide targeted AI consulting services and solutions, and truly bring AI into various industries.

A common saying in the AI ​​circle is that you cannot "look for nails with a hammer", but you must first find the nails and then match the hammer with the corresponding solutions and technologies to hammer through the pain points.

2. Find good talents first, then develop core competitiveness.

If you find a nail but don't have a hammer, you can only stare blankly. For technology companies, technological innovation is always the first competitive advantage. Who makes the powerful technological "hammer"? Of course, it's people.

AI is an intelligence-intensive industry, and the global competition for talent is fierce. In particular, my country is in a critical period of filling the gap in the field of independent and controllable technology, and must first find good talents to build the core competitiveness of science and technology companies.

Nortel Digital Intelligence is well aware of this and regards talent as its "No. 1 project". Despite its short history, it has quickly built a diversified, full-level, multi-polar new talent ecosystem, bringing together top technical talents in computing power, algorithms, and data, led by Professor Dou Dejing, a famous artificial intelligence expert, as well as industry insiders in strategy, brand, market, and intellectual property.

(Dou Dejing, Chief Scientist of Nortel Digital Intelligence)

This new talent team not only enables the major industrial research projects undertaken by Nortel Digital Intelligence to be completed on time and at the right pace, but also forms a commercial closed loop with close cooperation between the strategy, market and product teams, accelerating the conversion rate of technology products, and every scientific research project will be transformed into a product. The four major AI products mentioned above: Spark Intelligent Computing, Qianjin AI Heterogeneous Computing Platform, Baota Large Model Adaptation Platform, and Red Lake Trusted Data Space are all formed in this self-circulating virtuous closed loop.

With a team of new-type talents with innovative spirit and mission drive, Nortel Digital Intelligence can better shoulder the responsibilities of a state-owned enterprise and demonstrate its independent innovation capabilities in key areas.

3. Find friends first, then create integrated technology solutions.

Full-stack AI service providers are the general trend and what the industry needs. The ecological and open nature of "Spark Intelligent Computing" can well bring together AI industry partners to jointly create integrated solutions.

Why are industry partners and developers willing to trust Nortel Digital Intelligence? Yang Zhen mentioned that Nortel Digital Intelligence does not pursue exclusivity. We believe that for the AI ​​industry to develop rapidly, we need everyone to unite together to find a breakthrough path faster. Therefore, we must be autonomous and controllable, and we must also open our arms to open cooperation and unite all outstanding leading companies to jointly explore and expand the future blueprint of AI.

Integrating knowledge and action, Nortel Digital Intelligence has established cooperation with domestic large-scale model manufacturers such as Baichuan Intelligence in the software ecosystem, jointly exploring the application of large-scale models in multiple industries such as government affairs, medical care, culture, education, manufacturing and robotics, and creating leading scenario-based AI solutions. In terms of hardware ecology, Nortel Digital Intelligence, together with nearly 20 GPU partners and AI deployment optimization technology providers such as Qingmao Intelligence, has filled the gap in full-stack AI software and hardware service capabilities, providing industry users with the most cost-effective hardware solutions.

Only by understanding the industry, having talents and friends, and putting long-term value first, can we establish such a full-stack AI methodology with sustainable competitiveness.

As a young company, why does Nortel Digital Intelligence have such a vision and determination?

According to Yang Zhen, as an AI native enterprise under Beijing Electric Control, the establishment of Beijing Electric Digital Intelligence is in line with the industrial trend brought about by generative artificial intelligence. It is also a major layout for Beijing Electric Control to adhere to its original mission of "serving the country through industry" and "four aspects", and to be based in Beijing, a highland of scientific and technological innovation, and directly point to the category of new quality productivity.

Mission is the compass of business. It is the original mission that drives Nortel Digital Intelligence to focus on long-term and core value pursuits and make long-term plans.

After comprehensively considering multiple factors such as national needs, user needs, partner needs, and its own development needs, Nortel Digital Intelligence broke the deadlock and established a systematic AI capability building, ultimately drawing a panoramic picture of AI that enables the industry.

I believe that in the future, driven by full-stack AI service providers represented by Nortel Digital Intelligence, all walks of life will continue to cultivate and release new quality productivity, stimulate the unlimited potential of the digital economy, and ultimately turn the blueprint of Digital China into reality.