A year after the explosion of large model technology, although no real killer applications have emerged, a direction widely predicted is increasingly becoming the AI entry point in the mouths of many big shots. It is the AI agent. Microsoft founder Bill Gates believes that AI will completely change the way users use computers. AI agents will become the next platform, just like Android, Apple OS, and Windows in the past.
On a PC equipped with an AI agent, users will no longer need to open separate applications. Instead, they will hand over all interactions to an intelligent assistant driven by a large language model. It understands all our personal preferences, knows all personal data, and we rely on it to call applications and meet all needs. Therefore, every PC will become a "privately customized" online assistant for users.
The best carrier for AI agents is still the PC.
From the explosion of large models last year to the "emergence" of Sora this year, the industry's enthusiasm for AI applications continues to rise. Some industry insiders believe that artificial intelligence must not be separated from these paths to the application of the ground: specifically, it includes three major pillars of intelligent devices, intelligent infrastructure, intelligent solutions, and services. It is not difficult to predict that the disruptive changes triggered by artificial intelligence have been widely implemented in various industries. However, for the general public, the most perceptible is still the terminal devices around them.
Advertisement
In fact, at present, various terminal devices have corresponding business layouts among manufacturers. For example, mobile phone manufacturers are all working hard to promote AI large models into smartphones, new energy vehicle manufacturers are all working hard to promote large models "on the car", and even some smart home brands are also trying to implement large models in smart homes. In the view of industry insiders, the best carrier for AI agents is still the PC.
Firstly, compared to the mobile terminal, the computing power of the PC terminal is greater, which is enough to support the corresponding large model computing power needs. TOPS is a commonly used indicator to measure AI acceleration performance, including the comprehensive AI computing power of CPUs, GPUs, NPUs, etc. At present, the computing power of a smartwatch is less than 10 TOPS, the computing power of a smartphone is about 10 to 30 TOPS, the computing power of a computer is about 30-200 TOPS, and the computing power of a smart car with actual L3 and below autonomous driving capabilities is currently about 100-400 TOPS. The computing power of an AI PC, which is also a smart brain, is currently about 100 TOPS, and it can reach 200-500 TOPS in the short term. In this way, it is a reality that individuals have 1000 TOPS of computing power.
Overall, the conditions on the PC side are better than on the mobile side, and the ecology of cars and smart homes is more mature, so the application conditions on the PC side are better at present.
Secondly, strong intelligence is mainly based on large computing power devices such as AI PCs, especially when users need to keep data locally to protect security and privacy, and some scenarios require offline. Whether it is the needs of individual deep knowledge bases or intelligent management settings, they are shaping the basic form of AI PCs. From the perspective of demand gradient, the computing power intelligence devices needed by super individuals are presented in three forms: Book (notebook), BOX (smart box), and Station (workstation).
For different users, their computing power needs are different. GPT smart guests, music smart guests, and avatar smart guests may be enough to meet general needs. LoRA smart guests, program smart guests, video smart guests, design smart guests, and e-commerce smart guests often need the level of AI PCs in the box. Scientific research smart guests can basically meet the needs of daily analysis of Station-level AI PCs, but further needs require machine groups, clusters, or call cloud computing power to meet needs. However, no matter what type of user, AI PCs based on strong computing power can solve problems better.Lenovo's Radical Advancement
In the exploration process of AI PCs, Lenovo was the first to clearly propose the definition of edge-side large models and personal intelligent agents. Among the existing PC market players, it became the first to shout "the one who eats crabs," and in terms of specific layout, it also appears more aggressive than others. As early as December 26th last year, Lenovo announced a key part of its AI PC concept - the progress of personal intelligent agents. Lenovo released the personal intelligent agent "Xiao Le" to developers and launched a recruitment plan for intelligent agent app developers.
From the proposal of the concept to the detailed plan implementation, it is not difficult to see Lenovo's eager "expectations" for AI intelligent agents. The reason why Lenovo is so aggressively promoting the implementation of AI intelligent agents is also due to the situation.
On the one hand, Lenovo has reached the "ceiling" of the traditional PC market, and its core business is facing impact. Lenovo urgently needs to reshape the competitiveness of its core business. Looking at the financial reports, Lenovo has been the top seller in the global PC market since the fiscal year of 2018, but its PC business revenue has shown a significant decline. In the third quarter of the fiscal year 2023, Lenovo's revenue was 111.83 billion yuan, with a year-on-year increase of only 3%, and the net profit was 2.394 billion yuan, a year-on-year decline of 24%. The smart device business group, composed of personal computers, tablet computers, smartphones, and other intelligent devices, saw a year-on-year decline of 13.7% in revenue.
Comparing with the same industry, Lenovo's own sales gross margin is also behind the industry competitors. Before the fiscal year of 2020, Lenovo's gross margin was less than 15%, and it is currently only 17%. In contrast, other PC manufacturers such as HP (HPQ.NYSE) and Dell Technologies (DELL.NYSE) have maintained a sales gross margin of more than 20% for many years, while Huawei and Apple's sales gross margins have remained above 40%.
The main reason lies in Lenovo's PC strategy, which focuses on occupying a larger market share and mainly provides more mid-to-low-priced products, resulting in a lower overall gross margin. Manufacturers like HP and Dell Technologies have mainly focused on high-gross-margin niche markets in recent years, such as high-end business laptops or gaming PCs. However, for Lenovo, which has been in the mid-to-low-end market for many years, upgrading to high-end is not as simple as just raising the price. It must have "solid" technological support, and AI PC-driven intelligent computers can just enhance the added value of its computers and play an important role in helping it enter the high-end market and increase gross margins.
On the other hand, Lenovo's business based on cloud services and other intelligent solutions can also follow the price increase of AI PCs to increase the overall revenue level. In recent years, the decline in terminal business revenue, led by PCs, has led Lenovo to increase its investment in the business composed of servers and AI servers, edge servers, and storage servers. At present, AI has become one of the most important businesses for Lenovo. In June 2023, Lenovo proposed to increase investment by $1 billion in the next three years to accelerate the deployment of AI in global enterprises.
Lenovo's continuous investment and promotion in AI PCs not only directly bring the price of its PCs to a level above 15,000 yuan, taking the high-end route, but also promote its own AI servers and expand B-end business.
Huawei Focuses on Balance
Compared with Lenovo, Huawei's current AI core is still the smartphone, and the PC is just an entry point for its multi-end collaboration. In Huawei's view, the focus of AI PCs lies in multi-end collaborative capabilities, which is closely related to Huawei's own terminal product structure.Firstly, in Huawei's terminal products, there are not only PCs but also smartphones, smart cars, wearable devices, data clouds, and more. Therefore, in Huawei's AI strategy, the role of PCs is still focused on collaboration. Among Huawei's terminal products, smartphones hold a higher status, and they remain the core port for AI applications. In addition, the smart car series "Aito" that Huawei collaborates with is also an important battlefield for carrying AI applications. Thus, compared to Lenovo, Huawei's attitude towards PCs is more inclined towards multi-terminal collaboration. By leveraging Huawei's unified HarmonyOS system, it connects numerous usage scenarios, links many terminals, and creates new scenarios, thereby driving the growth of Huawei's terminal business revenue. In fact, this strategy of Huawei has been quite successful, with the terminal business sales revenue reaching 251.5 billion yuan in 2023, a year-on-year increase of 17.3%.
Secondly, based on the role of PCs in AI, Huawei has focused on brand-new technical upgrades, ecosystem expansion, and product renewals around PC products to enhance their AI PC capabilities. At the HarmonyOS ecosystem Spring Festival communication meeting held by Huawei, they launched the first PC product based on the PanGu AI large model, the Huawei MateBook X Pro. As an ultra-lightweight OLED notebook weighing only 980 grams, it is based on Huawei's powerful computing foundation and intelligent interconnection capabilities, supporting "AI Summary," which can quickly and accurately extract key information from text, audio, and video content.
In addition, the MateBook X Pro is also equipped with Huawei's AI Space feature, integrating partner applications such as WPS AI, Wenxin Yiyan, iFLYTEK Xinghuo, Zhi Pu Qingyan, Wondershare Filmora, and Lingjing Matrix. Through AI Space and hundreds of intelligent agents, along with some AI applications, Huawei is trying to lead a new trend in AI PCs.
Will personal intelligent entities bring about an OS-level intelligent revolution?
At present, in the matter of AI PCs, both Lenovo, which lives by selling PCs, and Huawei, which is always active on the forefront of technology, are actively preparing for this potential new opportunity. So, will the personal intelligent entity, seen as the first entry point, bring about an OS-level intelligent revolution? The answer may not be clear yet, but the trend seems to have emerged.
On one hand, the personalized experience brought by AI intelligent entities has a complete opportunity to bring personalized experiences to PC consumers, and manufacturers who create this kind of personalized experience will have the opportunity to gain a competitive advantage. Currently, even notebooks equipped with AI assistants do not have much stickiness for consumer-end users, and they are more at the level of being fun and useful. The introduction of personal intelligent entities allows users to find that calling application functions through personal intelligent entities is superior to directly entering fragmented applications for operation. At this time, the user's dependence on terminals with personal intelligent entities will be greatly enhanced.
Industry insiders have said that under the new AI wave, the new OS operating system will be the service form of AI large models, personal intelligent entities, local knowledge base calls, and AI native application connections, which is a brand-new OS upgrade. At present, manufacturers including Huawei and Lenovo are trying to do similar things. Once its potential is verified, it will eventually enter multiple terminals. At that time, on the computer side, mobile phone side, tablet side, and Internet of Things device side, there may be more and more personal intelligent entities, and manufacturers who can achieve cross-terminal integration may also be in an advantageous position in such competition.
On the other hand, although the entire industry is talking about AI PCs, the exploration of AI PCs is still in the early stage, and there is no clear profit model. Although the entire industry is talking about AI PCs, there are not many products on the market, and consumers' understanding of AI PCs is only that the computer CPU performance of AI PCs is higher, and their AI capabilities are still very vague.
From the developer's perspective, the industry's profit model is not yet clear, and all aspects still need to be explored by the entire industry. For example, the issue of application deployment, OEM PC pre-binding is just one way, but it is definitely not the only way. However, from the early AI PC pre-installed universal chat robots, to more technology combined with local and personal resources, and then to the arrival of some applications, the progress of AI PCs is visible to the naked eye. For example, the RAG retrieval enhancement generation technology has almost become a standard in a very short time, which can well illustrate the problem.
In this regard, AI PCs do indeed have new opportunities hidden in them. But for all manufacturers, this is destined to be a process of "feeling the stone to cross the river." Therefore, in the process of exploring AI PCs, all participating manufacturers should be prepared on both hands: on the one hand, actively track market reactions, and find out the real needs of the market in time; on the other hand, quickly follow up with technical trends and the latest business models to form a closed-loop commercialization plan, so as to quickly run through the commercial path and pave the way for the potential trend of AI PCs.