The first media of the industrial internet. Industry Family
If we envision from the perspective of the transformation of human-computer interaction, the changes brought by AI PCs can be said to be another “disruption” in the history of human-computer interaction. In the era of the metaverse, it was Apple that integrated AI into the Vision Pro; so in today’s AI era, who will completely bring AI PCs into reality? Who will be the ultimate winner?Author|Si HangEditor|Pi YeProduced by|Industry Family
ENIAC——the world’s first electronic computer, was programmed by six women during World War II. At that time, all women, including them, were not allowed to enter the ENIAC room. Looking back at history from today’s modern society, the status of women at that time presented an extreme state—these six women’s names could not appear on the “merit list” for a long time, despite their significant contributions to modern programming.
The ENIAC, source: history.com
And this was the first instance of human-computer interaction in history.
In the following half-century, from the “graphical user interface” revolution led by Xerox to the “touch interaction interface” revolution led by Apple, every transformation of human-computer interaction in history has stirred up great waves in the global PC market. This evolutionary process not only brings practicality, entertainment, and convenience to users but also carries a greater vision of a digital “utopian” world.
And today’s “AI revolution” is the vehicle that makes this vision a reality someday in the future. For example, today’s computers are just tools, while future computers could be super assistants. They can replace users in completing basic tasks and even creative work by storing users’ personal habits and knowledge when using the computer. More importantly, the privacy of such AI computers is no longer a problem.
However, before this utopian “AI computer” arrives, several questions worth discussing are: How to integrate AI into personal computers? How can personal computers perform large-scale AI computations? What requirements does this pose for hardware such as chips? Beyond hardware, what changes will the software layer undergo?
As a new wave of human-computer interaction led by AI reignites the PC market, a series of questions are becoming the focus on the PC stage.
1
How Capable Are Edge Large Models?
From smartphone cameras to smart wearables, from cars to smart homes, every input and output point may become a stage for AI to showcase its capabilities.
In early 2023, at the MWC World Mobile Communication Conference held in Barcelona, Honor launched products such as Magic OS 8.0, Magic Large Model, and Magic6 Pro smartphone; Qualcomm and MediaTek showcased chips with more AI capabilities. In the same year, Samsung, vivo, OPPO, Honor, Huawei, and Xiaomi began to apply large model capabilities to smartphones, accelerating the penetration of AI smartphones.
Thus, a revolution in AI occurring at the edge officially sounded the alarm. First smartphones, then smart wearables, and next cars and smart homes; it seems that all terminals can “add” AI.
So, why are various terminal scenarios combining with AI?
From the development trend of large models, advancing from cloud to edge is one of the current key development directions.If all deployment methods are cloud-based, it requires access to the cloud through terminals. Although this method can ensure sufficient parameter volume and computing power, it evidently lacks advantages for lightweight, low-latency tasks.
The core of edge AI lies in its ability to quickly respond to user needs and process local data without relying on cloud servers, thus protecting user privacy.
However, deploying large models at the edge still poses some challenges. The most obvious challenge is how to ensure that edge large models have enough computing power to support them?
From the current market’s edge large models, a reasonable rule is: the larger the device (the more functions it has), the larger the parameter volume of its edge large model.
In February 2024, Mianbi Intelligence, in collaboration with Tsinghua NLP Laboratory, released and open-sourced the edge large model Mianbi MiniCPM, with a parameter scale of 2 billion;
In September 2023, Xiaomi’s AI large model was unveiled, with a parameter scale of 1.3 billion;
In January 2023, Honor released the edge platform-level AI large model—Magic Large Model, with a parameter scale of 7 billion.
The aforementioned models deployed on smartphones show that their parameter scales cannot compare with the currently leading large models both domestically and internationally. For smart wearable devices with more limited functions, their edge large model parameter scales will only be smaller than those of smartphones. However, it is worth noting that Honor’s Snapdragon 8 is equipped with the Magic Large Model with a parameter scale of 7 billion.
In past dialogues with the media, Deng Bin, the president of Honor’s R&D management department, also revealed, “Running large models at the edge is limited by computing power, bandwidth, and power consumption; running a 7 billion parameter model at the edge has reached the limit.”
However, at the recent 2024 Beijing Auto Show, SenseTime’s “Riri New 5.0” can be considered a bombshell. It adopts a mixed expert architecture (MOE) and is the first large model in China to fully benchmark and even surpass GPT-4 Turbo, with a parameter scale of 600 billion. This is a more suitable edge large model for vehicle deployment.
In fact, whether it is Honor’s Magic Large Model or SenseTime’s Riri New, edge large models share a characteristic: large models with significant parameters deployed at the edge will definitely undergo “distillation” technology to compress the large model before being packaged at the edge.
Although large models will begin to “emerge” as their parameter volumes increase, and present Scaling Law (scale effect), this is also why everyone is competing for parameters; but from the perspective of model application, the parameter volume of the model is not necessarily better when it is larger. The ideal state should be: to achieve the best effect with as few parameters as possible.
Therefore, large models packaged at the edge need to undergo “distillation” processing. For example, Google’s MobileBERT model significantly reduced the model parameter volume through knowledge distillation and other technologies, making it more suitable for deployment on mobile devices.
Additionally, besides “distillation” processing, evolving from smartphones to PCs also requires chips that can ensure the normal operation of edge large models. In fact, over the past year and a half, many chip manufacturers have launched chips suitable for AI PCs.
In October 2023, Qualcomm launched a new Arm architecture processor designed for PCs: Snapdragon X Elite;
On December 14, 2023, Intel released the Meteor Lake Core Ultra processor, which is a CPU integrated with an AI acceleration engine NPU;
In April 2024, AMD launched AI chips suitable for commercial laptops and desktops—the Ryzen PRO processor.
Not only chips, since Intel launched the “AI PC Acceleration Plan” last October, the entire industrial chain’s ecology is beginning to reveal its shape. From independent hardware suppliers to independent software vendors (ISVs), as the underlying chips and upper-layer AI PCs “take their positions,” some hardware and software compatibility issues will also accelerate resolution.
However, compatibility adaptation is not easy. The hardware configurations of different devices vary greatly, and the diversity of operating systems and development environments requires that large models must possess high flexibility and portability. Moreover, large models deployed on PCs also need to ensure user privacy, which requires solving privacy issues based on effectively utilizing local data for model adaptive learning.
2
Who Are the Pioneers?
“We believe AI PCs are a key turning point in the PC market in the coming years.” In September 2023, at the “Innovation 2023” summit held in San Jose, California, Intel’s CEO Gelsinger first proposed the concept of AI PC.
This was during Intel’s large chip launch event. While showcasing its third-generation AI chip roadmap, Intel also emphasized the upcoming Core Ultra processor “Meteor Lake,” which Intel regards as the most significant processor architecture change in 40 years and a chip capable of being mounted on AI PCs.
This processor is expected to officially debut in December 2023. In fact, as a chip giant, Intel’s progress in AI PCs has always started from the hardware level, providing underlying support for AI applications through integrating AI accelerators and optimizing CPU architectures.
As early as 2018, Intel began laying out AI PCs. At that time, the “Athena Project” was its attempt in AI PCs, especially in optimizing battery management, voice recognition, and security. For instance, Intel’s 11th generation Core processor integrates a GNA unit specifically designed for low-power AI processing.
Meanwhile, in the international market, Microsoft is an undeniable force. As a leader in the operating system field, Microsoft integrated AI elements into its Windows 10 system as early as 2019, such as the Cortana intelligent assistant and machine learning-driven performance optimization features.
“Our goal is to enable every Windows PC to benefit from the efficiency gains brought by AI,” said Microsoft CEO Satya Nadella.
In the layout of AI PCs, Microsoft focuses on the intelligence at the operating system level, continuously deepening Cortana’s functionality through Windows updates, while utilizing AI to optimize system performance and user experience.
To welcome the “AI PC Year”, Microsoft also launched the Surface Pro 10 commercial version and Surface Laptop 6 commercial version in March this year.
Finally, leading the way in the AI PC market is the domestic PC manufacturer “Lenovo.” In December 2023, Lenovo first released AI Ready AI PC products, including the ThinkPad X1 Carbon AI and Lenovo Xiaoxin Pro 16 AI Core version. These products are equipped with Intel’s Core Ultra processor and an AI-exclusive NPU chip to provide strong local mixed AI computing power.
“Lenovo’s AI PC will go through the AI Ready phase and enter the AI On phase in 2024, at which point everyone will have their personal AI assistant,”said Abulikemu Abulimit, Vice President of Lenovo Group, at the conference.
Immediately, in April 2024, Lenovo released new AI PC products, and the “Lenovo Xiaotian” featuring a personalized AI intelligent agent officially debuted.
In fact, Lenovo’s vision for AI PCs is to create them as personal AI assistants, providing personalized creation, private secretary, and device management services. The local mixed AI computing power of Lenovo’s AI PCs is one of its core features, equipped with embedded personal intelligent agents for multimodal natural language interaction, as well as personal large models and local knowledge bases.
As early as 2020, Lenovo launched the Yoga series AI laptops equipped with Lenovo’s intelligent engine LCE, aiming to enhance user experience through AI technology, such as intelligent heat dissipation adjustment and scene recognition.
It can be seen that in the AI PC team, Lenovo belongs to the “pragmatic faction”, focusing more on the practical application of AI in real-world scenarios, achieving intelligent optimization in heat dissipation, battery life, interaction, and more through its intelligent engine LCE.
Additionally, besides Lenovo, Microsoft, Intel, and other internationally renowned manufacturers, there is another PC manufacturer that cannot be overlooked—Apple. Just before the publication of this article, Apple held its annual launch event, where the most stunning product was the AI-focused M4 chip.
Therefore, from the overall market structure, on one hand, brands like Lenovo, Microsoft, and Intel are leveraging their influence and technical accumulation to dominate. Although AI PCs are currently in the market cultivation phase, manufacturers are also exploring paths suitable for themselves.
On the other hand, emerging forces like Huawei and Xiaomi are attempting to cross into the AI PC market by leveraging their AI experience in the smartphone sector, forming a new competitive landscape through the integration of software and hardware advantages.
At the model level, manufacturers are primarily providing lightweight large models for developers, such as Google, Mianbi Intelligence, and Meta.
Data source: each company’s official website, CICC Research Department
However, in the future, with the technology of model compression and distillation, more complex large models are expected to run efficiently on PCs, such as OpenAI’s GPT series, Alibaba Cloud’s Tongyi Qianwen, Tencent’s Hun Yuan, Baidu’s Wenxin series, etc., all of which could become important components of edge AI PCs.
In fact, the compression technology of large models has also been validated in the market. For instance, Lenovo has compressed LLM to a lightweight model for local deployment based on large model compression technology. Currently, the Lenovo AI Now assistant’s large model comes from Alibaba Cloud’s Tongyi Qianwen, compressed from the original large model of 14.4GB to 4GB, which can run on computers with 5-6GB of memory.
As the industrial chain matures, the AI PC era is accelerating. IDC predicts that the penetration rate of AIPC is expected to reach 85% by 2027.
It is foreseeable that in the current trend of AI becoming the era’s trend, this AI-led transformation is also igniting the PC market.
3
Future Prospects of AI PC
The term “AI PC Year” is mentioned more and more frequently, but this track still lacks a “killer application”.
In September last year, at the Lenovo Innovation Technology Conference, while launching the “AI PC Acceleration Plan”, Gelsinger called on ISVs, “Want to know how killer applications are born? The answer is just missing you. It requires you to jointly shape the AI PC ecosystem and create killer applications.”
Indeed, killer applications are like the fuse that can ignite the industry for AI PCs. In the metaverse era, Apple’s Vision Pro became the catalyst; and in today’s AI PC era, it relies on the perfection of the entire ecosystem.
Moreover, to integrate large models into PCs, the real issues to be solved include model compression technology, while also ensuring hardware performance keeps pace, as well as how to ensure privacy security and software-hardware compatibility.
First, Microsoft views AI as the core of future operating systems, planning to evolve Windows to make AI PCs personal assistants and productivity tools for users. Currently, Microsoft is investing in advanced machine learning models, such as for more efficient data processing, personalized services, and strengthening cloud collaboration.
Similarly, Lenovo also mentions “personal assistants”. Among them, creating personal AI assistants to provide personalized services is Lenovo’s strategic roadmap for AI PCs. Additionally, in terms of ecology, Lenovo is also cooperating with software and hardware manufacturers through the “AI PC Acceleration Plan” to gain an advantageous position.
As a chip manufacturer, Intel is pushing the progress of AI PCs through hardware innovation.In this regard, Intel also plans to integrate more powerful AI acceleration modules into the next generation of CPUs, such as more efficient neural computing sticks, to provide low-latency, high-energy-efficient computing capabilities for AI PCs. Intel’s vision is to make every PC a node for edge computing, achieving local data analysis and instant feedback while collaborating with the cloud to provide users with a seamless AI experience.
Finally, as an emerging force in the AI PC field, HP, Huawei, and Xiaomi are participating in the market transformation of AI PCs from the perspectives of AI enterprise solutions, AI chips, and smart home IoT.
Additionally, on the hardware side, to adapt to the AI PC market, the upstream chip side of the industrial chain needs to undergo industrial upgrades. For example, how to ensure that large models have enough computing power at the edge? This requires operating systems to meet the computing power demands in mobile scenarios, and issues such as battery consumption, smartphone storage, etc., are being addressed one by one.
According to Trendforce, Microsoft plans to set minimum thresholds for AIPC in Windows 12, requiring at least 40 TOPS of computing power and 16GB of memory. Additionally, structural changes have occurred on the chip side, such as architectural changes, heterogeneous computing, and memory upgrades.
With Intel, Lenovo, and other international manufacturers endowing AI PCs with new concepts, the entire PC market’s explosion will follow. Whether viewed from the perspective of enhancing productivity or from the broader context of the AI era, this round of AI transformation will undoubtedly ignite enthusiasm in the PC industry.
Looking back at the two years of rapid growth of large models in China, AI infrastructure has gradually matured, and PCs are the ultimate carriers for AI to land. The reason is simple, the discussion points about large models have shifted from the large models themselves to AI Agents, super assistants, and intelligent agents, etc., and the best carriers to achieve this leap are PCs.
If we envision from the perspective of the transformation of human-computer interaction, the changes brought by AI PCs can be said to be another “disruption” in the history of human-computer interaction. In the metaverse era, it was Apple that integrated AI into the Vision Pro; so in today’s AI era, who will completely bring AI PCs into reality? Who will be the ultimate winner?
Video number content recommendation▼▼Industry AI Large Model《Financial Large Model, Where is the Industry Heading?》《The Boiling Large Model: Who is Taking the Lead?》《The Large Model Landing in the Financial Industry, Where is the Imagination? | Industry Feature》《Reinterpreting Baidu Intelligent Cloud: Written in the 24 Hours After the Large Model Opened》《A Southern School, Encountering AI for 60 Days》《Kingdee, “Dancing” in the Era of Large Models》Industry Digitalization《Under the Trend of Digitalization, the “Tightening Spell” of Sci-Tech Innovation Board IPOs《透视京东,和隐藏起来的200个上升密码》《Behind the Cloud Management Car: Insights into the Internal Strengths of Enterprises in the Digital Era》《Insights into Cloud Testing: Technology, Industry, and the Digital War under Cost Reduction and Efficiency Enhancement Industry SaaS《2021年SaaS行业八大趋势|盘点》《美团阿里鏖战，餐饮 SaaS无「终局」》
《Kingdee Teng “Cloud”, Industry Interconnection “Networking”》
《Tencent Leverages, SaaS Launches Team Battle
Industry Supply Chain《In the Era of Industrial Internet, Who is Rewriting the Supply Chain? 》《Joining Hands with State-Owned Assets, Looking Again at Suning’s Retail Industry Boundaries Industry Hard Technology《Cloud from “Knocking on the Door” Sci-Tech Innovation Board》《The Internet of Things, a Giant Battle for the Cloud》《Huawei: A New Story Told with HarmonyOS 2.0》《Domestic Chip Battlefield: Constraints, Determination, and the Visible Future of the Next Decade》© Previous Review▼
Industry Family Website Newly Launched© THE END
This article is produced by Industry Family Original, please do not reprint without permission. /Welcome to Report▼Industry Family Report Contact WeChat: 15201450756