What is the value of AI Ether?

Investment Strategy

In terms of network connectivity, Ethernet is actively embracing AI application scenarios. NVIDIA recently launched SpectrumX-800 Ethernet, and it is expected that its Ethernet business will contribute billions of dollars in revenue in the next fiscal year, demonstrating the immense potential of Ethernet in the AI field.

Looking to the future, Ether is expected to become the mainstream AI network interconnection solution:

1. The Ethernet has better scalability and openness for large-scale clusters.

2. As the development of AIGC accelerates, there is a greater demand for cluster capabilities and computing power.

3. The advantages of Ethernet in terms of cost and performance are gradually becoming apparent.

Currently, AI Ethernet has mainly formed two major camps: the Super Ether Alliance (UEC) and NVIDIA Spectrum-X. Whichever solution ultimately dominates will promote the expansion of the switch market space and accelerate industry development.

In terms of optical modules, the LPO( linear driver plug-and-play optical module ) solution may become the best choice for Nvidia's AI Ethernet. LPO has advantages such as low power consumption, low latency, and low cost, which help continue to implement the system's low-cost and low-power characteristics. The penetration rate of LPO is expected to accelerate, but it requires higher collaboration between upstream and downstream manufacturers, giving leading companies a greater advantage.

Suggestions to pay attention to:

• Switch industry chain: from chips, manufacturing to brand owners are expected to benefit

• Optical module leaders: such as Zhongji Xuchuang, Xinyi Sheng, etc.

• Liquid cooling, cloud computing power and other AI-related fields

Industry Trends

1. The global top 10 list of optical modules has been released, with Chinese manufacturers performing remarkably. Xuchuang Technology ranks first for the first time, with a total of 7 Chinese manufacturers on the list.

2. Google announced a $2 billion investment in Malaysia to build a data center and develop AI.

3. Micron expects HBM memory sales to reach billions of dollars in fiscal year 2025.

4. T-Mobile acquires US Cellular's wireless business for $4.4 billion.

5. Canalys predicts that the AI smartphone market share will reach 54% by 2028.

6. Gartner predicts that global public cloud spending will exceed $675 billion in 2024.

7. The country's "Big Fund" Phase III has officially been established, with a registered capital of 344 billion yuan.

Risk Warning

The risks include AI development not meeting expectations, computing power demand not meeting expectations, and intensified market competition.