Summary of Key Points
NVIDIA is preparing for the GTC Taipei conference by heralding a "new era for PCs," with plans to launch the N1X chip based on the Arm architecture (targeted at its flagship AIPC products), competing directly with Apple's M5 Pro/Max. Analyst Ming-Chi Kuo suggests that while the N1X could potentially replace Macs, it is currently aimed at a niche market (with annual shipments of around 10 million units). The success of this chip depends on whether the Windows ecosystem can provide robust end-side AI applications. Meanwhile, manufacturers like Qualcomm are accelerating their efforts in developing AIPC products, and the ecosystem is gradually taking shape. However, the current demand for end-side AI in the PC market is not yet rigid, but sectors such as storage and cooling within the supply chain are expected to benefit.
Detailed Analysis
#### 1. NVIDIA's N1X Chip: How Powerful Is the "Super Brain" for Flagship AIPCs?
The N1X is an Arm-based processor designed specifically for high-end AI PCs (AIPCs), with impressive specifications:
- Process and Performance: It utilizes TSMC's 3nm manufacturing process, which is more power-efficient and offers higher computational power. The chip features a 20-core CPU (10 high-performance cores + 10 low-power cores) combined with a Blackwell architecture GPU (48 computing units, 6144 CUDA cores), providing an end-side AI performance of 200 TOPS (200 trillion AI operations per second), sufficient for running large models locally.
- Memory and Target Audience: It supports 128GB of unified memory (shared by both the CPU and GPU for faster data transfer) and is aimed at heavy AI users, such as designers and programmers, directly competing with Apple's M5 Pro/Max, with the goal of capturing Apple's high-end market.
- Background: NVIDIA has been making progress in this area for years, starting with CUDA acceleration for creative software and moving on to integrating Tensor Core capabilities in RTX GPUs for local AI tasks like super-resolution and noise reduction, gradually bringing cloud computing power to PCs.
#### 2. Ming-Chi Kuo's Cautionary Note: N1X Remains Niche, with Windows Being the Key Factor
As a seasoned industry analyst, Kuo highlights two key points:
- Niche Market Status: Annual PC shipments amount to around 200 million units globally, so N1X devices are expected to sell only about 10 million units in the next two years, primarily targeting advanced users who need to run large models locally (e.g., researchers working with sensitive data).
- Success Depends on Windows: The key to widespread adoption is whether Windows can effectively integrate end-side AI capabilities. Currently, most AI applications on PCs are standalone and cannot cross-applicationally integrate data (e.g., combining information from emails, documents, and spreadsheets).
- Lack of Rigid Demand: Most users still rely on the cloud for AI tasks (e.g., using ChatGPT through browsers), so there is no immediate need for local computing power. The success of the MacBook Neo lies in its price and design, not in its end-side AI capabilities. Similarly, inexpensive mini-computers can also utilize cloud-based AI services.
#### 3. Device-Side AI vs. Cloud-Based AI: Which Is Better for You?
The main difference between the two approaches lies in where the data is processed:
- Device-Side AI: Advantages include privacy (data remains on the device, reducing the risk of leaks) and the ability to deeply integrate local data (e.g., work documents, photos). Disadvantages include higher costs due to the need for powerful hardware and reliance on operating system support for cross-application functionality.
- Cloud-Based AI: Advantages include no requirement for high-performance hardware and accessibility anytime. However, there are privacy concerns as data is transmitted to servers, and it relies on internet connectivity.
For most users, cloud-based solutions are sufficient for now, so the demand for device-side AI has not yet become essential.
#### 4. The AIPC Ecosystem: A Competition Among NVIDIA, Qualcomm, and Apple
Leading manufacturers are competing in the AIPC market with different strategies:
- NVIDIA: Focusing on the high-end market with the N1X, targeting heavy AI users and directly competing with Apple.
- Qualcomm: Taking a more pragmatic approach with its Snapdragon X platform (45 TOPS of computing power). It already has 85 Windows AIPC products available from brands like Lenovo and Dell, and plans to reduce the price to $600 (about RMB 4,300) in 2026 to target the mid-range market. The platform supports over 750 applications, including "Fortnite," dispelling concerns about the poor performance of Arm-based ecosystems. Its market share goal has been lowered from 30% to 12%, with a focus on long-lasting business laptops.
- Apple: Winning through its ecosystem and design. The success of the MacBook Neo is due to its price and user experience, not necessarily because of its end-side AI capabilities.
#### 5. Opportunities in the Supply Chain
The proliferation of AIPCs will drive growth in various sectors:
- Storage: Running large models requires more memory. For example, the Llama2 model requires 42GB of memory, and Lenovo's AIPC comes with 32GB of RAM and 1TB of flash storage, benefiting manufacturers like Micron and Samsung.
- Cooling: Increased computing power generates more heat, leading to demand for advanced cooling solutions (heat sinks, graphite sheets), providing opportunities for companies like Phoronix.
- PCB Circuit Boards: Higher demands for low-power, high-density (HDI) PCB manufacturing will drive orders for related manufacturers.
- Structural Components: There is a trend towards lighter and more premium designs (e.g., using metal or carbon fiber casings), benefiting companies like Coson Technology.
Conclusion
AIPCs represent the future of PCs, but they are currently in a niche phase. For widespread adoption, Windows needs to develop a robust end-side AI ecosystem that makes local AI more convenient for ordinary users. Meanwhile, sectors in the supply chain, such as storage and cooling, are already poised to benefit from this trend.