Value soared by 30%, AI PC drives semiconductor shipment surge | AI Dehydration

Author: Zhang Yifan

Editor: Shen Siqi

Source: Hard AI

The Taipei International Computer Show is set to grandly open on June 2nd.

As the exhibition approaches, various phenomenal AI PCs are also gearing up.

Just last week, Lenovo, at its performance meeting, first proposed their "Five Elements" definition of AI PCs, sparking heated discussions both inside and outside the industry.

Shortly thereafter, Morgan Stanley also provided their unique insights and definition of AI PCs. Morgan Stanley believes that a true AI PC should have the following two core features:

1. Equipped with a dedicated Neural Processing Unit (NPU)

2. High processing speed (exceeding 40 TOPS)

Moreover, Morgan Stanley boldly predicted that the popularization of AI PCs will bring unprecedented gains to the entire industry.

1. AI PCs Driving Industry Growth

Morgan Stanley predicts that with the upgrade of processors and memory, the semiconductor value of each AI PC will increase by 20%-30%, and the average selling price of PCs will also rise by 7%.

The main difference between AI PCs and traditional PCs lies in the addition of a dedicated Neural Processing Unit (NPU). Traditional PCs use a CPU+GPU architecture, while AI PCs adopt a CPU+GPU+NPU architecture.

To fully leverage the NPU's performance, AI PCs are also equipped with larger memory capacity. Microsoft pointed out that basic AI models require 16GB of memory, standard AI models require 32GB, and advanced AI models require 64GB or more.

Based on this, Morgan Stanley predicts:

• Due to the upgrade of processors and DRAM, it is expected that the semiconductor value in each AI PC will increase by 20-30%, bringing an additional $30 billion in revenue to global semiconductor companies;

• More advanced components (such as NPU, more powerful CPUs, and larger memory) will raise the average selling price (ASP) of computers. Compared to the general market consensus, the ASP of computers is expected to increase by about 7%.

2. AI PC Architecture Trending Towards ARM

Morgan Stanley points out that in the era of AI PCs, as processors shift from x86 to ARM architecture, it is expected that by 2027, ARM architecture PC CPU sales will reach 50 million units, generating $10 billion in revenue.

x86 and ARM are two mainstream processor instruction sets.

• x86 architecture has long instructions and is good at expressing multiple tasks with a single complex instruction;

• ARM architecture has short instructions and is good at breaking down complex tasks into multiple instructions; The difference in expression makes the ARM architecture have the characteristics of high performance and low power consumption. Compared to x86, the ARM architecture is more suitable for high-power AI computing.

Currently, Qualcomm, NVDA, AMD, and MediaTek have successively launched AI PC processors based on ARM architecture.

Morgan Stanley pointed out that with the growth of the Windows on Arm (WoA) ecosystem, it is expected that by 2027, the sales of WoA PC chips will reach 50 million units, bringing in $10 billion in PC CPU revenue.

3. Competition of AI PC Processors

Morgan Stanley stated that Qualcomm's recent launch of the Snapdragon X Elite processor for AI PCs is worth paying attention to.

Since acquiring Nuvia, Qualcomm's processor performance has greatly improved. The latest architecture is also used in the released Snapdragon X Elite, which, according to Bank of America's report, leads among various AI PC processor manufacturers in terms of performance.

In the recent batch of AI PCs launched by Microsoft, the Qualcomm Snapdragon X Elite chip is used to support Copilot. These computers include the Surface Laptop and Surface Pro.

In addition, according to data compiled by Morgan Stanley, other major computer manufacturers such as Lenovo, Acer, ASUS, Dell, HP, and Samsung have also released their next-generation AI computers. These computers are also equipped with X Elite/X Plus chips (see the image below).

In the past, most of Microsoft's processors used Intel CPUs. According to market data, in Q4 2023, Intel's share in the PC CPU market was 78%. This share may change with Qualcomm's strong entry into the market.

4. Forecast of AI PC Shipment Volume

2024 is considered the first year of AI PC shipments, and Morgan Stanley has also provided a forecast for the shipment volume of AI PCs.

Morgan Stanley predicts:

• In 2024, AI PCs will account for 2% of the total computer shipments;

• By 2028, AI PCs will account for 65% of the total computer shipments;

This means that by 2028, the shipment volume of AI computers will reach 179 million units, with 85% being laptops.

Thanks to the shipment of AI PCs, Morgan Stanley points out that component suppliers, computer manufacturers, and supply chain participants will benefit:

• Component suppliers: Companies like Nvidia, Qualcomm, AMD, and Intel, as key suppliers of NPU and CPU for artificial intelligence computers, are expected to reap huge profits.

• Computer manufacturers: The report points out that Lenovo, Dell, Asus, and HP will be the major beneficiaries as they hold advantageous positions in the computer market.

• Supply chain participants: With the increasing demand for components for artificial intelligence computers, companies such as SK Hynix (memory), MediaTek (chipsets), Delta Electronics, and Advantech are expected to benefit. Leading chip foundry TSMC is also expected to benefit from the increase in CPU orders.