Source: Liu Kuang WeChat Official Account

Image Source: Generated by Wujie AI

As expected, the wave of AI large models has reached the field of smartphones. Smartphone manufacturers are racing to develop their own AI large models or collaborate with AI large model manufacturers to explore the huge potential of integrating smartphones with AI large models.

On October 26, Honor CEO Zhao Ming announced some information about the new Honor Magic6. He stated that the new device will be equipped with the new Snapdragon 8Gen3 and Honor's self-developed 7B-side AI large model (7B refers to a scale of 70 billion parameters).

On October 16, vivo announced the release of a self-developed AI large model matrix, including five self-developed large models with different parameter levels of billions, tens of billions, and hundreds of billions, covering core application scenarios. These large models will be first applied in vivo's new operating system, OriginOS 4.

On October 11, OPPO announced the official start of the beta version 1.0 of the new Xiao Bu assistant, built based on its self-trained AndesGPT (AndesGPT) large model. The upgraded Xiao Bu assistant will have AI large model capabilities.

Earlier, Xiaomi Chairman Lei Jun announced that Xiaomi is fully embracing AI large models and stated that large models on the mobile side have been preliminarily tested. Xiao Ai, the voice assistant, has also been upgraded with AI large models and has started invitation testing. Similarly, Huawei announced the integration of the Pangu large model into its mobile system, initiating public testing of the built-in large model voice assistant Xiaoyi, and released the new Huawei Mate60 series with built-in AI large models.

It can be foreseen that large model smartphones will become increasingly popular in the future and become the mainstream trend in the smartphone market.

Collaboration Between Large Model Manufacturers and Smartphone Manufacturers

The changing demands of users, the support of technological advancements, and the pressure of market competition all urge smartphone manufacturers to develop large model smartphones. Large model manufacturers also intend to expand into more industries to implement large models in various industries.

For AI large model manufacturers, the transition from general large models to industry-specific large models is a necessary step for technological implementation. General large models have wide applicability but lack industry depth, and they can hardly provide high-value services in vertical fields. Therefore, large model manufacturers actively implement large models in hardware fields such as smartphones and smart learning machines.

Firstly, the training and inference processes of general large models consume a significant amount of resources, resulting in huge operating costs. In contrast, large models used for smart hardware have fewer model parameters, lower costs for training, debugging, optimization, and implementation. Secondly, large model manufacturers provide end-to-end solutions, including hardware, software, and algorithms, making it more convenient for customers to use and deploy AI models. Thirdly, entering the smart hardware field can help AI large model manufacturers expand their market share and increase their competitive advantage.

For smartphone manufacturers, the lack of innovation in smartphone products and a significant decline in the entire industry chain have prompted them to embrace large model technology to enhance the image recognition and processing capabilities of smartphones, strengthen voice interaction capabilities, support auxiliary functions and intelligent assistance, thereby improving user experience and enhancing enterprise competitiveness.

AI large models demonstrate highly accurate and widely applicable capabilities in natural language processing, image recognition, and speech recognition. Integrating them into smartphone systems can enhance smartphone performance. Furthermore, with the evolution of AI algorithms, large model smartphones can provide personalized services based on user data analysis, achieving a "one-to-one" relationship.

On one hand, AI large model manufacturers are actively entering the hardware field (smartphones, smart learning machines) to meet the increasing demand for intelligence. On the other hand, smartphone manufacturers are intensively embracing AI large models to improve smartphone performance and enhance differentiation advantages. With the joint efforts of large model manufacturers and smartphone manufacturers, the exploration of large model smartphones has gradually transitioned from the stage of technological research and development to the stage of application and industrial layout.

In conclusion, the consensus of smartphone manufacturers such as Huawei and Xiaomi is to bring large models to smartphones. However, they have different ideas and progress in terms of strategic layout and implementation paths. This new "war" of large model smartphones is quite promising.

Huawei's "Leading" Position

While other smartphone manufacturers are still exploring the path for the implementation of large models, Huawei has already released large model smartphones, taking the lead in the "large model smartphone era."

Huawei's ability to rapidly and systematically develop and produce large model smartphones is attributed to its long-term strategic vision, strong R&D team, efficient production processes, excellent project management, and continuous innovation and iteration. These factors support each other, enabling Huawei to maintain a leading position in the market competition.

Technologically, Huawei focuses on both high-end smartphone technology and AI large model technology, providing technical support for the development of large model smartphones. In terms of smartphone manufacturing technology, Huawei's self-developed technologies such as smartphone chips and operating systems have a domestication rate of over 90%. The Huawei Mate60 series is a strong external representation of its technological capabilities.

In terms of large models, Huawei's Pangu large model includes Pangu NLP large model, CV large model, multimodal large model, and scientific computing large model. It must be said that Huawei's large model smartphones are gaining more attention. By closely following emerging technologies and continuously investing in mastering these new technologies, Huawei adjusts its related strategies according to market demands.

From the full integration of the Pangu large model into HarmonyOS 4, becoming the world's first mobile terminal operating system embedded with AI large model capabilities, to the appearance of the flagship smartphone with the "smart image search" function based on multimodal large model technology, Huawei has been at the forefront of technological innovation.

In terms of products, Huawei is the first domestic smartphone manufacturer to integrate large model capabilities into its smart assistant (Huawei Xiaoyi) and land it for consumers. Huawei Xiaoyi integrates Pangu's natural language large model, visual large model, and multimodal large model, with the highest version reaching 100 billion parameters. It started public testing in September this year, and the first batch of supported models is the Mate 60 series smartphones.

Subsequently, at the Huawei Autumn Full-Scene New Product Launch, Yu Chengdong also revealed, "After more than a month since the release of HarmonyOS 4.0 with large model support, the number of upgraded users has exceeded 60 million, making it the fastest-upgraded version of HarmonyOS in history."

Technological and product capabilities are important resources for the development of large model smartphones. In terms of smartphone technology and large model technology, Huawei has deep technical accumulation and strength, enabling it to provide large model smartphones with outstanding performance and enhance its competitive advantage in the field of large model smartphones.

Of course, the label of "pride of domestic products" is also one of the reasons why Huawei's large model smartphones can be quickly accepted and recognized by consumers. A good brand image can establish consumers' trust in Huawei's products and services, differentiate it from other "friendly competitors," attract more consumers, and help it lead in market share.

Xiaomi's "Gradual" Progress

Like Huawei, Xiaomi is also advancing into the field of large model smartphones. Although its progress in deploying large models for smart phone terminals is not fast, its offensive is strong, and it has already achieved some results.

It is understood that Xiaomi's self-developed 1.3 billion parameter edge model has been successfully tested on mobile phones, and the effects in some scenarios can rival the computational effects of the 6 billion model in the cloud. In addition, in terms of data, Xiaomi's self-mined and organized training data accounts for 80%, with Xiaomi's own product and business data reaching 3TB.

On October 26, at the Xiaomi Surge OS and Xiaomi 14 series new product launch event, Lei Jun also announced, "Xiaomi's group strategy will be upgraded from 'smartphone X AIoT' to 'people, cars, and home full ecology.' He emphasized that the new strategy will be centered on people, organically integrating the people, cars, and home full ecology, and the key to the new strategy is Xiaomi Surge OS."

It is worth noting that Xiaomi Surge OS is one of the largest-scale bottom-layer restructured operating systems in history, and it has deeply integrated AI large models. This enables the Xiao Ai input assistant to have text creation and AI expansion capabilities, which are reflected in the Xiaomi 14 series.

Thus, Xiaomi has achieved a phased victory in the exploration of large models and has revealed its attitude of "aggressively attacking" large models.

On one hand, Xiaomi announced full embrace of large models, actively forming an AI large model team, and continuously making large-scale R&D investments. Such a high degree of emphasis on talent reserves and technological investment is beneficial for enhancing Xiaomi's innovation capabilities, technological leadership, risk reduction, and progress in the field of large models.

In April, Xiaomi officially established an AI Laboratory large model team, with over 1,200 personnel in the AI field, and the head of the team is Lu Jian, head of the AI Laboratory large model team of the Xiaomi Technical Committee. Before this, Xiaomi's AI-related teams had established capabilities in AI technologies such as vision, speech, acoustics, knowledge graph, NLP, machine learning, and multimodal.

In terms of R&D investment, Xiaomi adheres to a "technology-oriented" approach and spares no effort to "spend money" to achieve technological innovation. Financial report data shows that in 2022, Xiaomi's R&D investment was 16.2 billion yuan, and this year's R&D investment is expected to exceed 20 billion yuan. In the next five years, Xiaomi's total R&D investment is expected to exceed 100 billion yuan.

On the other hand, Xiaomi has clearly defined its R&D direction, focusing on the "lightweight" and "local deployment" of large model technology, striving for the differentiation of solutions for large models on smartphones, and creating unique product advantages.

As is well known, from the perspectives of multitasking demands, network speed and cloud computing development, changes in application ecosystems, user needs, and market trends, the trend of large model smartphones integrating the edge and the cloud is inevitable. It involves user experience, user privacy, and is a key point for smartphone manufacturers to layout the field of large model smartphones.

In this regard, Xiaomi chooses to combine the edge and the cloud, with the edge as the main and the cloud as the auxiliary. It deploys large models to the edge so that information does not need to be uploaded to the cloud, and all computations are done locally, fundamentally ensuring that user privacy is not compromised. Additionally, Xiaomi adjusts the model structure and parameter size to adapt to the characteristics of various chips in terms of memory and computing power, achieving the best integration of large models with hardware products.

In summary, the new competitive landscape forces Xiaomi to strengthen its differentiation strategy in terms of products, technology, etc., in order to gain a larger market share in the fiercely competitive field of large model smartphones.

Impact on the Next Growth Frontier

From the current market situation, it is an undeniable fact that the smartphone market is in a state of weakness. According to the statistics from the Chao Dian Think Tank, the total sales volume of the top 20 best-selling smartphones globally in Q2 2023 was 145 million units, a 7.6% decrease compared to 157 million units in Q2 2022. Overall, the smartphone market in 2023 is still in a state of continuous decline.

The continued weakness of the smartphone market makes industry innovation a breakthrough key, and large model technology may be the "elixir" for rejuvenation.

Firstly, the rapid development of AI large model technology is expected to drive continuous improvement in the functionality and performance of smartphones, attracting a new generation of consumers to upgrade their devices. Empowered by AI large model technology, intelligent assistants become more powerful and intelligent, and smartphone cameras gain strong image recognition capabilities.

Secondly, smartphone manufacturers such as Huawei and Xiaomi actively integrating AI large models into product features or production processes is of great significance for driving the development of the smartphone industry.

Large model smartphones not only bring better experiences and convenience to users, but also enhance product performance and efficiency, while also creating business opportunities and competitive advantages for smartphone manufacturers. However, most players are still in the research and development stage, and there are not many truly mass-produced large model smartphone products. In the future, smartphone manufacturers will continue to face many challenges that need to be addressed.

Furthermore, the development of AI large model technology is still in its early stages, and there is still a lot of innovation space for large model smartphones in the future. With the advancement of AI large model technology, the value of smartphones is expected to be redefined, and we may see more advanced and intelligent smartphones, thereby stimulating market vitality and accelerating the recovery of the consumer smartphone industry.

All signs indicate that large model smartphones are the next frontier of industry growth, sparking a new round of technological competition among smartphone manufacturers. In the future, more smartphone manufacturers will compete to launch more innovative and functional products to meet user needs, leading the development of smartphone technology, and selecting a group of winners with strong technological capabilities and innovation.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。