Report Producer:Shenwan Hongyuan
The following is an excerpt from the original report.
1.AI boosts the new cycle of PCs.
After the stay-at-home economy, AI innovation drives the recovery of the PC cycle.
From a cyclical perspective, the PC market will recover in 2024. After bottoming out and recovering in 2019, the global PC market has experienced a continuous high boom from 2020 to 2021 due to demand for remote work during the epidemic**. According to IDC data, the total shipment volume of the global PC market in 2022 will be 292.3 billion units, a year-on-year decrease of 165%。
Since 2022, PC has gone through 2 years of inventory clearance, and although the 2023H2 peak season is not prosperous, it has now recovered to a healthy inventory level of 6-8 weeks. Lenovo, HP, Dell, Acer, Asus and other five Wintel brand manufacturers are looking forward to PC shipments in 2024.
From the perspective of innovation, the global PC industry will enter the AI era in 2027. Qunzhi Consulting**, AI CPU and Windows 12 will be released in 2024, which will become the first year of large-scale shipment of AI PC. It is estimated that the global shipment of AI PCs will reach about 13 million units in 2024 and become a mainstream category in 2027.
2.Intel and Qualcomm push AI CPUs
NB brand and chip competition landscape.
According to IDC data, the top 5 PC brand markets in 2022:
1) Lenovo, with shipments of 68 million units and a market share of 233%
2) HP, with shipments of 55.3 million units and a market share of 189%
3) Dell, with shipments of 49.8 million units and a market share of 17%.
4) Apple, with shipments of 28.6 million units and a market share of 98%
5) Asus, with shipments of 20.6 million units and a market share of 7%.
The NB processor market is highly competitive, and AMD and Apple continue to gain Intel market share. According to IDC data, Apple's M-series chips currently have a market share of about 10%, AMD Ryzen processors and AMD Radeon graphics cards are strongly combined, and Intel is facing market competition and the failure of the "Moore's Law" of semiconductor manufacturing processes, and its customer computing division (CCG) performance is under pressure.
Andy-Bill's law rotates to promote the landing of AI PCs.
The penetration of generative AI in terminals (PC + mobile phones) has deduced a new version of "Andy Beale" law. In ICT, there is a rotational iterative relationship between "AI application-terminal-chip communication". End-to-end AI is the first in PCs and mobile phones, followed by end-side AI in wearables and smart homes.
Through the integration of software and hardware, AI PCs will decentralize AI functions that would otherwise be available in the cloud to personal PCs, reducing data transmission delays, information security, and privacy issues.
Lenovo's AI PC concept.
Lenovo AI Territory - "Three Large Models" Framework. In addition to public and private domain models, local devices such as mobile phones or PCs can achieve new dimensions of innovation with the help of personal models. Yang Yuanqing, chairman and CEO of Lenovo: Let everyone have their own large model.
End-side model: Identifies and evaluates the importance of all associated substructures in the network, and then prunes and quantifies them. The significance of device-side large models and AI applications: 1) Make full use of edge computing resources to reduce dependence on cloud computing resources. 2) According to the user's Xi and other data, under the premise of ensuring privacy and data security, form a personalized PC mobile phone use experience.
AI PC Hardware Implementation Platform:
Intel + Lenovo: 30% of innovative ideas are accelerating on Meteor Lake + 70% of innovative ideas will be used on future PC platforms. Lenovo said at Tech World 2023 that the AI PC will be available around September 2024.
Google develops the integrated advantages of device-side AI.
Google: It has an integrated layout from chip to AI application.
Gemini is the first model to outperform human experts in MMLU (Massive Multitasking Language Understanding).
End-to-end hardware: Based on the self-developed Tensorg3 chip, the AI mobile phone Pixel 8 series was launched, which can run Google's AI basic large model, and Tensorg3 is equipped with RIO TPU to enhance the computing power of AI inference and other aspects.
AI application: Assistant with Bard has the AIGC function, as an office assistant on the mobile phone, assisting in the operation of DOCS, Gmail and other applications.
Apple's A and M series chips have built-in NPUs
A-series SoC: Built-in 0. since A116 tops NPU, 35 tops of A17 computing power in 2023.
M-series SoC NPU performance iteration: 11 tops 158 tops→18 tops(fp16)。M3, M3 Pro, and M3 Max have an enhanced neural engine that accelerates ML models.
Intel builds GPU+NPU+CPU** computing power for AI PCs.
The GPU, NPU, CPU** computing power of Intel Ultra series processors.
The CPU consists of three computing cores of different specifications: P-Core (Performance Core @cpu module), E-Core (Energy Efficiency Core @cpu module), and LP E-Core (Low Power Efficiency Core @soc module).
Arc GPU can provide powerful computing power support for game rendering, processing, and AI computing.
A low-power AI engine for NPU for continuous AI and offloading AI loads, such as sound processing and image processing in meetings.
Meteor Lake: Advanced manufacturing process, chiplet and 3D advanced packaging.
The Core Ultra processor Meteor Lake is seen by Intel as the most significant change in the client SoC architecture in 40 years:
For the first time, a chiplet is used to integrate an NPU in an SoC module for AI local inference acceleration
With Intel 4 process node and 3D high-performance hybrid architecture, the energy consumption ratio has been significantly improved. Intel 4 transistors are twice as dense as Intel7 and 20% more energy efficient.
For the first time, FoverOS 3D heterogeneous packaging technology (CPU + SoC + GPU + IO 4 tile) was adopted.
Intel Core Ultra supports real-time AI applications.
Intel believes that AI PCs should have four characteristics:
1) Eliminate network latency and provide instantaneous response AI services2) The lowest cost of computing power and the easiest to use;
3) Highly personalized;4) Strong personal privacy protection, sensitive data does not need to leave the local area.
In terms of AI application scenarios, the Ultra series chips all support the AI effect of calls, and the highest-end series Ultra 9 can support AI-enhanced editing.
Taking the flagship Core Ultra 7 165H as an example, compared with the Core i7-1370P, with the blessing of the new CPU and GPU, the processing performance of UL Procyon Video Editing is increased by 31%, the processing performance of Premiere Pro is increased by 55%, and the processing performance of Lightroom is increased by 10%.
Intel and 100+ ISVs promote the AI PC application ecosystem.
Intel: The industry's first AIPC acceleration program.
In September 2023, Kissinger first proposed the concept of "AI PC". "AI PC is a key turning point for the PC market in the coming years. "We expect to ship tens of millions of new AI-enabled PCs to the market by 2024, and then expand to hundreds of millions. ”
Leveraging the PC processor industry's broadest ISV ecosystem, Intel is unlocking a range of new applications for AI PCs:
Optimized for AI PCs with 100+ AI ISV partners
300+ AI-accelerated ISV functions in 2024;
100 million+ AI processors in 2025;
Intel press conference: 23 well-known domestic ISV software vendors showcased AI applications jointly created with Intel.
Comparison of mainstream chips for AI PC.
At present, Qualcomm, Intel, and AMD all have chips for AI PCs, and Apple's M3 series supports hardware acceleration functions.
AI PCs such as Lenovo and Asus were the first to be released.
AI PC product launch: Lenovo, Acer, Asus, and others have officially announced new products equipped with Intel's new processors. Intel expects to supply chips for more than 230 AI PCs from partners such as Acer, Asus, Dell, DynaBook, Gigabyte, HP, Lenovo, LG, Microsoft, MSI and Samsung.
Lenovo Group and IDC jointly released the industry's first "AI PC Industry (China)**", emphasizing that the development of AI PC is a continuous evolution process, which should be divided into AI ready stage and AI ON stage.
In the AI Ready stage, AI PC is mainly an upgrade of the chip computing architecture, with basic local hybrid AI computing power, which can provide basic guarantee for software and service innovation of AI PC and start experience innovation.
By mixing AI computing power, the ThinkPad X1 Carbon AI can give users nearly 8 times more AI energy efficiency. Optimize software applications: Zoom, Microsoft Teams, Studio Effects, Adobe Lightroom, and more for features like text**, background blur, meeting notes, email summaries, and more. Lenovo AI PC will be the first to carry a personal large model in 2024, and the combination of AI and PC will form a new hybrid of "computing power platform + personal large model + AI application".
End of excerpt from the report For more information, please read the original report
List of topics in the report collection x Regularly compiled and updated by [Report School].
(Special note: This article ** is public information, the excerpt is for reference only, does not constitute any investment advice, if you need to use, please refer to the original report.) ）
Technology, Electronics, Semiconductors
Artificial Intelligence |AI Industry |AI Chip |Smart Home |Smart Speaker |Intelligent Voice |Smart Home Appliances |Smart Lighting |Smart Toilet |Smart Terminal |Smart Door Locks |Smartphone |Wearables |Semiconductors |Chip Industry |3rd Generation Semiconductors |Bluetooth |Wafer |Power Semiconductors | 5g |GA RF | igbt | sic ga | sic gan |Discrete |Compound |Wafer |Packaging & Testing |Display | led | oled |LED package |LED Chip |LED Lighting |Flexible folding screen |Electronic components |Optoelectronics |Consumer Electronics |Electronic FPC |Board |Integrated Circuits |Metaverse |Blockchain |NFT Digital Collection |Virtual Currency |Bitcoin |Digital Currency |Asset Management |Insurance Industry |InsurTech |Property & Casualty Insurance |