来源 | 伯虎财经（bohuFN）
Author | An Xiao
On April 16, AMD launched AI chips for business laptops and desktops, Intel launched the Core Ultra chip at the end of last year, and plans to launch a new generation of Intel Core Ultra client processor family this year, and is expected to ship 40 million AI PCs a year.
As global chip manufacturers become more and more mature in AI technology and continue to launch a new generation of AI chips, PC manufacturers are also scrambling to bring their own AI PC products.
According to media sources, "Apple will launch new AI PC devices this year, and the first M4 AI chip is expected to be released in June." Among them, Lenovo held the 2024 Innovation and Technology Conference on April 18, bringing more than 10 new AI PC products, and released the industry's first AI PC personal agent "Lenovo Xiaotian".
In 2024, a battle for the AI PC market has begun.
01 2024 is expected to start the first year of AI PC
The so-called AI PC, which stands for Artificial Intelligence Personal Computer, is a new generation of products that apply artificial intelligence technology to personal computing devices. In September last year, Intel CEO Pat Gelsinger first proposed the concept of AI PCs in Silicon Valley. He believes that "an AI PC is a PC that can play with AI functions." ”
What's so amazing about a PC that can play with AI?
Different from ordinary PC devices, AI PCs can implement AI functions, including but not limited to AI voice dialogue, AI video enhancement, AI cutout, AI painting, and AI article summary. For consumers, AI PCs provide smarter and more personalized services, improve their work, learning efficiency and entertainment experience, and for the market, the emergence of AI PCs will help drive consumer demand.
Therefore, after the heavy blow of the epidemic, consumers are gradually returning to rational consumption, and global PC shipments have fallen for two consecutive years. The emergence of the AI PC concept is like a "catalyst", catalyzing the demand and development of the market.
This year is also expected to open the first year of AI PC.
First, in line with the development of the AI era, PC intelligence has entered the AI PC era.
(Source: Guotai Junan Securities)
According to the "2024 AI PC Industry Market Space, Competitive Landscape and Global Leading Enterprise Analysis Report", the development of PC intelligence has gone through four stages of development: "traditional PC", "PC intelligent exploration", "PC intelligent upgrade" and "AI PC era".
In 2022, with the rapid development of generative AI and LLM and the improvement of hardware computing power, AI application scenarios are becoming more and more abundant, and the technical foundation of AIPC is also becoming more and more perfect.
Second, during the trough period of global PC shipments, AI PCs injected new vitality into the industry.
For every PC manufacturer and business, no profit can be made sooner. According to IDC data, after the short dividend period brought by the epidemic, global PC shipments began to decline for several consecutive quarters in the fourth quarter of 2021, and experienced a decline in shipments for nearly two years.
The environment is like this, and the situation of the players in it can be imagined. For the nine months ended December 31, 2023, Lenovo's revenue was $43 billion, down 13% year-on-year, and attributable net profit was $763 million, down 49% year-over-year. Specifically, in the third quarter, Lenovo's revenue was $15.7 billion, up 3% year-on-year, and its net profit was $337 million, down 23% year-on-year.
(Source: IDC)
In Q1 2024, global PC shipments increased by 3% year-on-year, ending a two-year downward trend and returning the industry to pre-pandemic levels. This change is due to the advent of the AI PC era, the recovery of shipments in various fields, and the upgrade cycle of end users.
Based on this, many analysts believe that "2024 may be the first year of AI PC". Qunzhi Consulting predicts that with the release of AICPU and Windows 12, 2024 will become the first year of large-scale shipments of AI PCs, and IDC also predicts that by 2027, 6 out of every 10 PCs shipped will be AI PCs, with a compound annual growth rate of up to 62%.
It can be seen that PC manufacturers have to join this AI PC war, otherwise they will face the risk of falling behind in shipments.
02 What is the color of each product?
At present, the average lifespan of PCs is more than 5 years, and manufacturers must find ways to stimulate user consumption to shorten this cycle if they want to increase sales. Everyone is also here with AI PC logic, and I want to use AI to shorten this replacement cycle.
Judging from the above data, the emergence of AI PCs has indeed stimulated the growth of the industry, but the impact is not enough. In closer detail, this 3% growth is also due to the acceleration of PC manufacturers' new product launches.
(Source: IDC)
According to IDC, the combined share of six vendors rose to 78.9% from 77.4% last year, while the shipments of other vendors fell 5% year-on-year. It can be seen that only the leading PC manufacturers have maintained the trend of shipment growth, among which Lenovo ranks first in the world with 13.7 million units of shipments.
Last year, Lenovo continued to launch new products, including the ThinkPad T series, ThinkPad P series, and ThinkPad X1 series, and in October of the same year, Lenovo also showcased its first AI PC concept machine and a series of native function applications.
In June 2023, Apple released the MacBook Air with the M2 chip, and its M3 series was also released at the end of 2023.
Intensive new product releases will obviously drive sales growth, which is common knowledge in the industry. However, the side also confirms that the effect of AI PC to stimulate the market may not be as good as everyone's expectations, which is related to the immaturity of AI PC technology and the lack of product landing.
What are the differences between AI PCs and past PCs? Yang Yuanqing said in an interview that each company has different definitions of AI PCs, some think that as long as there is an NPU (neural network processing unit), it is an AI PC, and some can use large cloud models and so on. In his view, AI PCs should have five characteristics: running AI agents that are compressed and optimized for performance and naturally interacting with users, having stronger computing power and supporting local heterogeneous computing power combining CPU, GPU, and NPU, accommodating more personal data throughout the life cycle and forming a personal knowledge base, an open AI application ecosystem, and protecting personal data and privacy.
At present, all AI PC products are still in the early stage, and they are exploring and trying to implement AI applications. Everyone still needs to prove their strength, and it is difficult to say that commercialization opens the market and expands sales, so as to achieve the purpose of shortening the user replacement cycle.
Specifically, from the perspective of each AI PC product, it is divided into two ways: self-developed AI chips and embedded AI chips and processors developed by others.
In China, Lenovo is the one with the largest force and voice in the field of AI PC.
In addition to the AI PC concept released last year, the company also jointly released the "AI PC Industry (China) White Paper" with IDC. At the Innovation and Technology Conference, more than 10 AI PC products were released in one go, covering Lenovo ThinkPad, ThinkBook, YOGA, Xiaoxin and other product lines, which made everyone really feel Lenovo's determination to "AI for All to fill the world with AI".
Previously, the AI concept in Lenovo's Pro 16 IMH, ThinkPad X1 Carbon AI and other AI PC products mainly focused on the hardware provided by Intel, including the Core Ultra processor and the NPU AI engine. This year, most of the newly released AI PC products are based on AMD processors.
Glory also adopts this way of pushing new products. On March 22, Honor's first AI PC product, the MagicBook Pro16, was officially launched, which is equipped with an Intel Core UItra7 processor and an NVIDIA GeForce RTX40 series GPU.
Overseas, Apple has adopted the first approach, with its self-developed M3 chip embedded in the MacBook Air series of products, and the company claims that the computer is "the best AI consumer notebook". In addition, industry insiders also predict that "Apple will launch the M4 series of chips in October this year." Presumably, Apple's new generation of AI PC products equipped with this chip will also arrive.
Up to now, Samsung, Dell, HP, Asus and other overseas PC manufacturers have started selling AI PC products or released related product plans. However, most of them rely on accessing other AI chips or processors to launch new products.
Regardless of whether it is an access chip or a self-developed chip, most of the current AI PC products only realize basic AI application functions such as AI voice assistant, face recognition, intelligent image processing, and automatic document analysis. For example, Lenovo's current AI PC products are equipped with self-developed Lenovo Xiaotian, which has built-in more than a dozen AI applications such as document summary, meeting minutes, AI painter, AI PPT, etc., to provide AI functional services for learning, work, social and life and other scenarios.
This kind of AI PC product that needs to combine a variety of AI applications obviously cannot be summarized by the concept of "PC that plays with AI functions", and this kind of product is not a small distance to realize the AI era of "intelligent Internet of Things" and "Internet of Everything".
It is precisely based on this that it is inevitable that there will be a saying in the market that "AI PC is still a pseudo concept and a fake product". This also reveals that everyone's AI PC product technology is not mature enough, and there is still a long way to go before commercialization.
03 It is necessary to cross the river by touching the "AI large model".
In this way, the real explosion period of AI PC may be far from coming. Not only domestic and foreign AI PC products are in their early stages.
The resistance behind this is not only due to technical limitations, but also to the fact that the computing power required is too high, the cost is high, and the implementation is difficult.
The essence of AI PC is the device-side deployment of large models, and the cost behind this is difficult to estimate. According to ICT Industry Watch, Zhou Jiangong, the founder of Unfinished Research, believes that AI PCs can truly solve user pain points by at least meeting the most basic conditions: computing power must reach 40 TOPS (trillion operations per second), model parameters must reach 70 billion, inference speed must be at least 20 tokens per second, context length must reach 4k tokens, and memory must reach 16G.
This also poses a cost problem for AI PCs.
Liu Jun, executive vice president and president of Lenovo China, once revealed, "The AI PC products launched in April are expected to be priced between $1,000 and $1,500, close to the price of 10,000 yuan, which is not friendly to most consumers." ”
At present, the price of Lenovo Xiaotian released by Lenovo is 5999 yuan, and the rest of the AI PC products equipped with this personal agent are priced at more than 10,000 yuan, of which ThinkPad T14p is priced at 14999 yuan, and Yoga Book 9i is priced at 17999 yuan.
In terms of the difficulty of landing, PC is also considered to be the "best carrier of large models", and the real integration of PC, large models, and AIGC also requires AI-specific computing power chips. The two are inextricably linked.
Nowadays, there are more than 100 AI models in the world, but there are few companies that break through, and even fewer have really changed user habits. The difficulty of landing large AI models can be seen, and the difficulty of device-side deployment is also even greater.
Moreover, on the consumer side, everyone's perception and acceptance of AI PCs is not high.
In the Baidu index, the overall daily average search volume for "AI PC" was 293, and the mobile daily average was 107, down 11% month-on-month and 17% month-on-month. It can be seen that AI PC is making a splash in the technology industry, but the actual search volume in China is not high, and it is showing a downward trend.
(Screenshot from: Baidu Index)
Nowadays, most consumers may not even know what an AI model is, let alone what an AI PC is, and they will not buy a computer for AI. It's like the concept of domestic robots appeared more than 80 years ago, but it has not yet been popularized in everyone's life.
After all, although the bigwigs are conclusive about the future of AGI, for the vast majority of people, the impact of AI on life is far less than that of short videos dominated by recommendation algorithms. This technology product is full of fantasy, but the investment is also huge, the cycle is long, and the return is uncertain.
Nowadays, AI has added a PC behind it, the scope of intelligence has been reduced, and the time is uncertain before it really lands into the public eye. But as a company on the wave, AI is not a choice. What will the next personal terminal look like? That's the definition they need to fight for.
Resources:
1. Huaxin Securities: Special Report on the Electronics Industry: AI Large Model Landing Terminal, AI PC Drives New Growth in the PC Industry
2. Guotai Junan Securities: 2024 AI PC Industry Market Space, Competitive Landscape and Global Leading Enterprise Analysis Report
3. DoNews: Lenovo doesn't want to lose, but AI PC is difficult to be a lifesaver
4. Smart Pro: PC out of the cold winter, AI has contributed a lot?
*The copyright of the first image and accompanying picture of the article belongs to the copyright owner. If the copyright owner believes that his work is not suitable for everyone to browse or should not be used for free, please contact us in time, and the platform will correct it immediately.