The Battle in the AI Midfield: Who Can Be the Best "Partner" for Enterprises?

Everyone has more or less had such a vision when imagining their ideal home.

Imagine this: When you come home tired from work, the air conditioner has automatically adjusted to the right temperature, and the floor cleaning robot has completed the cleaning during the day, leaving the floor spotless. When you walk into the kitchen, the smart refrigerator screen creates a dinner list for you based on the ingredients of the day. When you sit at the dining table, the TV turns on automatically and tunes to your favorite variety show. Before going to bed, you hear an audiobook playing from the speaker. When you feel sleepy, just say, "I'm going to sleep," and the whole house will automatically turn off the lights and shut down the unnecessary appliances...

Connecting all the devices in the home through a central brain to enable them to communicate and collaborate with each other in an orderly manner, to achieve an active and seamless fully smart home experience. As early as ten years ago, technology manufacturers represented by the previous generation of Internet of Things technology have made such attempts. However, due to the immaturity of interaction technology, the limited intelligence of devices, and the lack of a unified hardware ecosystem, the era of smart homes has been delayed.

Until today. The AI technology revolution represented by large models has swept the industry, bringing the feasibility of large-scale implementation. Intelligent productivity is gradually entering daily life, and the penetration rate of intelligent devices is increasing rapidly. According to IDC estimates, the connection rate of China's smart home device interconnection platform will be 76% in 2024, and the shipment volume will increase by 6.5% year-on-year; by 2025, the market size may exceed 830 billion yuan.

From October 11 to 13, 2024, at the "Mobile Home" booth of China Mobile's Global Partners Conference, with a mobile phone, linking six family scenarios such as AI technology, AI health, and AI learning, people's ideal vision of a smart home has been realized here.

Behind the integration and intercommunication of multi-scenario device, on the one hand, it is the change of people's demands for high intelligence and emotional consumption. The application innovation represented by China Mobile's "Mobile Home" will bring huge changes to specific fields. At the same time, this is also the unique path taken by China Mobile, working with partners to build an "AI+" ecosystem and release the industrial value of digital intelligence technology.

"All in AI" at the Conference

The audience's attention is still very focused on the booths related to AI. But an obvious trend change is that compared with last year's focus on showing the data processing and generation capabilities of the general large model, this year, more are the specific applications and implementation results of AI technology.

At the entrance of the "Mobile Home" booth, the most eye-catching is undoubtedly China Mobile's AI robot family, which consists of four home service robots: the humanoid biped robot "Fengqi", the wheel-arm robot "Xiaozhi", the wheeled companion robot "Xiaojia", and the quadruped robot dog "Xiaoli". Among them, "Fengqi" is exhibited for the first time and is planned to conduct offline pilots next year; the other three robots have made real machine demonstrations at the scene.

"Xiaozhi" is positioned as a personalized assistant in the pan-family field. At the booth, the staff placed several items of different shapes and materials on the table. When "Xiaozhi" received the cleaning task, it would first determine whether the item was recyclable or non-recyclable, and then extend its right arm dexterous hand and coordinate the two-finger gripper of the left arm to grasp the target and accurately put it into the corresponding trash can.

For the "Xiaojia" and "Xiaoli" robots, which have a faster R & D and commercialization progress, the scene also shared their roles as life assistants, family photographers, etc., helping people with tasks such as express delivery, vlog shooting, medication and call reminders, kitchen cleaning, meal preparation, etc., showing their capabilities in bionic interaction, intention understanding, task planning, and whole body coordination.

In the past few years, due to the immaturity of technology, robots could only provide case-by-case services, and the application space in complex scenarios such as families was limited. Until the emergence of large models, the development of embodied intelligence technology has greatly improved the cognitive and decision-making capabilities of robots, making large-scale implementation possible. For example, the above four AI robots are precisely a complex systematic project that China Mobile integrates network, computing power, storage, big data, and multi-modal technology based on years of data precipitation and intelligent algorithms.

The greatest imagination of AI lies in changing the physical world. If it is said that there is still a distance from the landmark leap of service robots into families, in industries such as industry, medical care, finance, and government affairs, the industrial revolution initiated by AI has already begun and has achieved remarkable results.

Take the industrial mine scene as an example. China is a vast country with abundant resources. Affected by different geographical environment variables, the coal mining industry faces challenges such as a shortage of personnel, difficulty in replication between mines, insufficient intelligence, and difficulty in scene implementation. A large number of manual on-site operations are still required 300 meters underground, and the main transportation system also relies heavily on manual inspections. Miners need to walk 10 - 20 kilometers for inspections every day. Not only is the workload large, but the environment is also extremely harsh, making it difficult to ensure safe and efficient operations.

After the introduction of China Mobile's Jiutian large model, based on its "1 industrial large model + 1 set of tool chain + 4 industrial AI applications" model, a service model dedicated to mines is developed in one - stop, using AI to assist and replace humans to complete complex and high - risk coal mine production work; its "AI + Industrial Safety Supervision" intelligent body, for the mine safety production supervision scene, with the three core capabilities of graphic and text understanding, comprehensive scheduling, and intelligent generation, realizes the intelligent upgrade of mine AI supervision.

AI, as a new generation of infrastructure, the industry consensus is that the intelligent era is not only the large model itself, but also the large model ecosystem. The native applications developed based on the large model have become the key for enterprises to break through.

At the event site, China Mobile released a new AI intelligent assistant "Lingxi". Based on China Mobile's Jiutian large model, "Lingxi" can not only provide basic services such as schedule planning, AI shorthand, and chat companionship, but also show the ability to solve problems in professional fields for scenarios such as communication, office, and home. For example, it can help users efficiently manage data such as contacts and information, and can also participate in family activities and control smart home devices.

It can be seen that at the Global Partners Conference, China Mobile has shown its various capabilities. It not only realizes large - scale applications in the home communication business but also focuses on a broader range of industries. It wants to work with partners to truly use AI to solve practical problems.

The Mid - game Competition of AI Lies in Applications

There is a high wall between AI technology and industry applications. Different from the development logic of the previous generation of technology industries where demand drives supply change, today's AI industry is "supply - demand" driven. Different industries and enterprises have their own problems, and the number of细分 scenarios is difficult to count. Many AI companies find it difficult to understand the differentiated needs of each link in special scenarios, and therefore cannot "prescribe the right medicine".

Even if the technical capabilities are transferred, it is inevitable to encounter lower - frequency, long - tail complex scenarios and the need for large - scale collaboration, so that some enterprises find it difficult to really use AI in their business. The characteristics of these customers are that they have a certain willingness to pay and a large number, and at the same time, they are very clear about their own needs. Therefore, they attach great importance to the professional performance of AI in the vertical field.

Take the medical industry as an example. In order to help doctors diagnose the condition, daily physical examinations are essential. In the past, the examination data was entered manually, which took a lot of time and cost. In response to this demand, China Mobile has launched a medical large model, which not only provides basic capabilities such as medical report generation, medical information extraction, and medical Q & A, but also cuts into the entire process of medical follow - up, consultation, and medical records, accelerating the landing of the large model in the real industry.

On the other hand, although the call to embrace AI is high, for traditional industries to use the technology, not all have a dedicated R & D team, and the cost of algorithm production is relatively high, making the input - output ratio difficult to guarantee. In the implementation process, enterprises often think about how to use a large model with less computing cost and higher intelligence level to provide AI services.

Looking at the three components of AI: the model is the largest arena where anyone can do something; data is the moat, and industries are like mountains; and computing power relies on a few giants.

Some industry insiders once analyzed to 36Kr that computing power is the power plant in the era of large models, and not all players can afford it. It is the general trend to give birth to several top operators. "Just like there is China Mobile in the communication network era, and there are Amazon, Alibaba, Tencent, etc. in the cloud computing era."

In this context, China Mobile, which focuses on AI + co - creation and pays attention to the basic base of digital intelligence services, coincides with the needs of enterprises for reliable AI partners.

According to the chairman of China Mobile disclosed at the event, since this year, China Mobile's two 10,000 - card - level intelligent computing centers in Hohhot and Harbin have been put into operation one after another. Next, in terms of facilities, China Mobile will also build a super 10,000 - card intelligent computing cluster, accumulate trillions of Tokens industry data sets, and train trillions of parameters of AI large models.

In addition, during the conference, China Mobile also launched a business customer computing power integrated machine with several partner enterprises, which is built based on the capabilities of network connection, Internet of Things, visual connection, and intelligent connection. Through the deep integration of network access and AI computing power, it provides support for the networking, monitoring, office, and security AI intelligent upgrade of small and medium - sized enterprises. 

Be the Partner of Enterprises in the AI 2.0 Era

Today, a new round of scientific and technological revolution and industrial transformation characterized by digital intelligence is developing in depth, and AI is accelerating its breakthrough and application. From the "Internet +", "5G +" to the "AI +" era, where will the technology ultimately land?

In the view of China Mobile, it will definitely be in thousands of industries and thousands of households. 

In order to maximize the release of AI capabilities to the application end, China Mobile has also proposed two parallel solutions: One is to rely on the base large model to build a多元 AI scene application matrix with the family as the core; the second is to serve partners, covering the entire chain from the base, computing power algorithm, model, terminal to application, to support different enterprises at various stages.

Specifically, first in terms of technology, China Mobile released the Jiutian Shanzhi Multi - modal Base Large Model at the Global Partners Conference, which is also the latest research and development result of the Jiutian series of general large models. According to the measured data, several key technical indicators of this model have reached the industry - leading level. 

In order to accelerate the large - scale application of the model in various industries and improve the ability to solve problems in professional fields, China Mobile has also深入 into more than 10 industries such as finance, transportation, energy, and manufacturing, and independently developed more than 30 industry large models spanning multiple fields, extending and exploring to more industrial levels, and accelerating the formation of an empowerment radiation circle for various fields of the economy and society. 

Secondly, on the product side, China Mobile has launched the fourth major customer brand "Mobile Home", building a smart home service ecosystem and service system of "one line + one network + one home", and ultimately realizing the ideal smart life of interconnection between people, smart devices and families.

Under the AI 2.0 wave, enterprises are undergoing disruptive industrial reengineering under the impact of large models. Ecology is the last mile to enter the AI era. Based on this thinking, the chairman of China Mobile also proposed four "promoting leaps" and four "model reconstructions" at the event site. 

First, "AI +" promotes the realization of leapfrog integration and innovation, and reshapes the economic and social development model; second, "AI +" promotes the realization of leapfrog industrial system, and reshapes the digital intelligence industry development model; third, "AI +" promotes the realization of leapfrog enterprise operation, and reshapes the enterprise value creation model; fourth, "AI +" promotes the realization of leapfrog information consumption, and reshapes the digital intelligence life service model. 

In the current situation of increasingly fierce international AI competition, multiple changes are an indisputable fact. A new ecosystem with AI as the core is forming, and technological innovation is leading the way forward. 

Adhering to open cooperation, mutual benefit and win - win results, and choosing a reliable partner will play a crucial role in how far an enterprise can go in the wave. China Mobile has obviously found its own position.