In 2025, AI will reshape the global PC market at a speed far exceeding people's imagination. This year, different vendors have presented new insights into AI PCs.

For the first time, edge-side AI approached the performance boundary of cloud-side AI when Apple released the M4 series of chips in March. The M4 Max chip, equipped in the top-tier Mac Studio model, leverages 128GB of unified memory and 546GB/s bandwidth to run the 200 billion-parameter Llama 3.1 large model locally and smoothly, achieving an inference speed of over 96 tokens/second (for the 8B model).

Enterprise-grade AI PCs are also beginning to chase profits. Intel launched the commercial vPro series of processors at MWC 2025; AMD's eight-card W7900/W7800 server solution based on the vllm framework provides small and medium-sized enterprises with a low-cost solution for deploying the DeepSeek R1 671B model.

The consumer market follows a strategy of "cost-effectiveness + full-scene collaboration." Xiaomi, Honor, and Lenovo have all launched their own AI PCs, driving AI PCs from "novelty items" to "necessities."

01 How has AI PC evolved in 2025?

According to the latest data from Canalys, global shipments of AI PCs reached 15.4 million units in the fourth quarter of 2024, accounting for 23% of total PC shipments for the quarter. PC shipments in mainland China reached 8.6 million units in the fourth quarter, of which AI PCs accounted for up to 20%. This share was only 7% in the fourth quarter of 2023, marking a triple increase in just one year.

Last year, AI PCs were still in their "enlightenment phase," utilizing local computing power on PCs to achieve applications such as generative AI poetry, text dialogue, and text-to-image generation. Similar to AI assistants, these functions were not very practical in actual use.

This year, the direction of AI PCs is clearer. AI PCs need to be able to assist people in multiple scenarios such as office work, entertainment, and healthcare through locally deployed applications.

For example, this year Honor's new AI PC strategy was upgraded to AI PC2.0, and the newly released MagicBook Pro 14 is touted as the "King of Tuning." It is adjustable from CPU, GPU, NPU, memory, voltage, fans, battery, to audio, focusing on comprehensive and in-depth tuning of AI software and hardware self-coordination. The CPU is equipped with

It is worth noting that Honor's new AI PC incorporates Chipsea Technology's EC (Embedded Controller) chip, the CSCE2010. This product is mainly developed for the mainstream consumer notebook market and can flexibly adapt to X86 and various domestic CPU platforms. The EC chip market has long been dominated by US and Taiwanese companies.

Xiaomi's AI PC product is the REDMI Book Pro16 2025. It is equipped with Intel's second-generation Core Ultra processor, featuring 16 cores and 24MB of cache. It integrates a new Intel Arc graphics card with an AI computing power of 96 TOPS and comes with 8400Mt of LPDDR5X ultra-high flash memory. In terms of AI, it can perform AI video content search functions and support AI document processing.

Apple's AI PC product is not a notebook but the Mac Studio. According to Apple, a Mac Studio equipped with the M3 Ultra can run large models with 600 billion parameters locally. It also adopts a unified memory mechanism, making it cheaper than deploying using NVIDIA GPUs and significantly reducing the demand for cooling.

Therefore, it is not surprising when a large model studio announced on Twitter that it successfully ran the 8-bit version of R1 at a speed of 20 tokens per second using two Mac Studios equipped with 512GB of M3 Ultra.

02 Who is paying for AI PCs?

From a market perspective, the country offers subsidies of up to 2,000 yuan (usually 15% of the selling price) for eligible AIPC products through the "trade-in" subsidy policy.

However, consumers' awareness of AI PCs is not sufficient. According to the "2024 AIPC Industry Trends and Potential Consumer Insight White Paper" jointly released by Tencent Marketing Insight (TMI) and GfK in October 2024, the report points out that the current awareness rate of potential consumers towards AIPC is low (59%), and their understanding of the AIPC concept remains vague.

Reporters visited multiple offline stores and learned that compared to the AI functions of computers, more consumers care about the configuration of the computers. A store clerk at a Lenovo store said, "Very few consumers pay for AI features, and computers can install Doubao, but very few users can clearly distinguish the functional differences between Doubao and Lenovo Xiaotian."

Of course, PC vendors will also actively demonstrate AI PC functions to consumers, but more users will perceive them as computer versions of smart assistants.

"Although the AI smart assistants that come with the vendors already support functions such as voice interaction, I still feel that cloud-based large models like Deepseek and Doubao are more powerful, and these large models can be directly installed and used on computers," a consumer who has experienced AI PCs told reporters.

Compared to the unclear desire for purchasing AI PCs on the consumer side, enterprises have a greater demand for AI PCs and are more willing to pay the cost.

On the one hand, the needs of enterprises are clearer. With the widespread application of AI technology in various industries, enterprises need to leverage the powerful computing capabilities of AI PCs to handle complex tasks. For example, financial enterprises need to conduct risk assessments and market forecasts through AI PCs; manufacturing enterprises need to utilize AI PCs to optimize production processes and quality control.

On the other hand, enterprises are also more willing to invest costs. Because AI PCs can bring enterprises higher production efficiency, better decision support, and stronger competitiveness. For example, enterprises can realize automated office processes through AI PCs, reducing labor costs; or by optimizing production processes, improving product quality and production efficiency.

03 What kind of AI PC can run DeepSeek

The official documentation of DeepSeek indicates that local deployment of the DeepSeek large model must strictly follow the three-tier resource matching principle of "video memory - memory - storage." This also means that if the CPU is added, focus must be placed on the four core hardware components.

According to the Hugging Face benchmark test, 4-bit quantization can compress the video memory usage of the 7B-parameter model of the DeepSeek-R1 version (hereinafter referred to as the DeepSeek-R1 version for simplicity) from 13GB to 4.2GB, but with a loss of about 8% accuracy.

Let's look at the hardware configuration required for the minimum 7B model:

The GPU requires an NVIDIA GTX 1080 (8GB of video memory) or a graphics card of equivalent performance. The graphics card needs to support FP16 computing and be compatible with NVIDIA CUDA models. The CPU requires 4 cores or more, such as Intel i5/i7 or AMD Ryzen 5/7 series processors. It needs 16GB of DDR4 memory. Memory is required to store a large amount of data such as model parameters and intermediate calculation results. 16GB of DDR4 memory can meet basic storage requirements and ensure the normal operation of the model. Storage requires at least 20GB of solid-state drive capacity.

To run models with 13B+ parameters smoothly, the hardware requirements are higher:

Since the size of video memory directly determines the scale of data that can be processed during model operation, the larger the video memory, the larger the model that can be run. Therefore, the GPU video memory should be 24G or more, such as NVIDIA RTX 3090/4090 graphics cards. The CPU should have 8 cores or more, such as Intel i9 or AMD Ryzen 9 series processors. Memory should start with 32GB of DDR5 memory. Compared to DDR4, DDR5 memory has significant improvements in speed and bandwidth. Storage requires a solid-state drive with a capacity of 1TB or more that supports the NVMe protocol.

Based on previous actual tests, computers from 6-7 years ago (such as those equipped with GTX 1060 graphics cards + 16GB of memory) could barely run 1.5B models with a generation speed of about 2-3 tokens/second; mainstream configurations from the past 3-4 years (such as RTX3060 + 32GB of memory) could smoothly run 7B/8B models (10+ tokens/second); 32B models require NVIDIA RTX30/40 series graphics cards (with video memory ≥ 12GB), but the generation speed is still limited by video memory bandwidth.

Looking at the current notebook options on the market, there are currently very few gaming notebooks to choose from, as the next cycle for gaming notebooks from major vendors will be tied to RTX50 series graphics cards, which are basically AIPC.

Currently, the highest configuration of the ROG Flow X16 2025 can locally deploy the 70B version of DeepSeek, with the highest option being AMD Ryzen AI Max+395, making it currently the most powerful AI-capable 2-in-1 notebook.

The clever aspect of this notebook is that it does not have a built-in dedicated graphics card but adopts a new integrated design that encapsulates high-performance GPU cores and high-speed memory into the CPU chip, supporting the local deployment of large-parameter AI models through large-capacity unified memory. The price is relatively expensive, with a pre-sale price of 20,999 yuan, reduced to 18,999 yuan after including the national subsidy.

Of course, in terms of inference performance and other aspects, there is still a gap compared to professional graphics cards, but it is already the lowest-cost solution for locally deploying AI models with a large number of parameters.

Currently, most notebooks still have some difficulties in running DeepSeek.

Acer's Aspire Go AI, priced at around 3,000 yuan after the national subsidy, is equipped with the Core Ultra 5-125H platform. Although it does not currently have an AI capability similar to Lenovo Xiaotian's "AI Assistant," Intel's native AI engine "AIGC" is compatible with all Ultra processors and also supports commonly used functions such as AI painting, document summarization, and report generation.

AI PCs priced at around 4,000 yuan include the HP Star Book Pro 14 AI, REDMIBook Pro 14 2024, and Honor MagicBook Pro 14 2025, which basically all come with AI assistants. HP has the "HuiXiaowei AI Assistant," Xiaomi has "Xiaoai Classmate," and Honor has YOYO Assistant, and they have officially accessed the DeepSeek large model. Compared to AIGC, the AI assistants built into these products have a wider range of applications, such as global callouts, webpage recognition, and document summarization, which are more convenient.

AI PCs priced at around 5,000 yuan after the national subsidy, such as the new ThinkBook 14+, use the Ultra 5-225H chip with a performance release of 70W. In terms of AI, Lenovo Xiaotian is the traditional vendor with the fastest progress in AI assistants and is likely the first traditional brand to announce full access to DeepSeek.

04 Worthy of Anticipation: HarmonyOS AI PCs

On March 14, it was reported that Microsoft's Windows system license for Huawei had expired that month and had not been extended. This means that Huawei's new laptops will no longer come pre-installed with Windows. Huawei insiders responded that the license given by the US Department of Commerce had expired and had not been extended.

Last September, Yu Chengdong revealed in a livestream that due to sanctions, "Windows PCs may stop being supplied, and this may be the last batch. In the future, we will only use our HarmonyOS PCs."

There are reports that Huawei is planning to launch five new laptop products, including the 2024 versions of the MateBook D 14, D 16, 14, GT 14, and X Pro. These new products will all use the Linux system. Due to the open-source nature of the Linux system, there is no need to pay additional licensing fees, which can reduce the overall cost of the computers.

There are also reports that Huawei's self-developed commercial AI notebook is expected to be released in April, equipped with the fully integrated DeepSeek large model, using Kunpeng CPUs and the HarmonyOS PC system for chips and operating systems.

Previously, Huawei's Qingyun series of commercial PCs achieved nearly 100% self-development rate, with processors, operating systems, and other core components all independently developed, building a full-stack controllable capability from hardware to software. In terms of AI PCs, Huawei has launched AI PC products equipped with local agents and edge-side large models. Taking the MateBook series as an example, it supports edge-side DeepSeek technology through the CPU+GPU+NPU heterogeneous computing architecture, realizing powerful "personal AI assistant" functions and significantly improving office efficiency.

Will Huawei's next PC be a HarmonyOS AI PC? It's worth looking forward to.