Ming-Chi Kuo: OpenAI ASIC AI server is expected to be shipped in the fourth quarter of 2026, and water-cooling design has become a new trend

Author: LoRA Time: 10 Jan 2025 883

Recently, well-known analyst Ming-Chi Kuo released an operational report on Wistron Technologies, a subsidiary of Wistron, revealing the latest developments in artificial intelligence servers. He pointed out that OpenAI’s upcoming ASIC AI server is expected to begin shipping as soon as the fourth quarter of 2026. The chip design of this server is carried out by Broadcom, a well-known technology company, and its technical strength cannot be underestimated.

According to Ming-Chi Kuo’s analysis, Amazon’s Trainium2 servers have begun shipping in large quantities in the air-cooled version, and Wiwynn’s production lines are operating at full capacity. It is worth noting that the water-cooled version of Trainium servers will also enter mass production in the fourth quarter of 2026, demonstrating the rising demand for higher performance and cooling technology.

Computer room data center server

In addition, Ming-Chi Kuo also mentioned that future ASIC AI servers will have higher computing power, which will cause the power consumption per cabinet to increase to more than 80–100kW. In order to cope with this demand, water cooling design has become an inevitable choice. Although the introduction of water cooling systems will significantly increase the cost of a single cabinet, its improved performance and efficiency will provide better services for the data center.

It is worth mentioning that Wiwynn Technology is currently re-discussing the relevant details of the GB200NVL72 order with Omniva, a major customer in the Middle East. Mass production is expected to begin in the second or third quarter of 2026, and the first batch of orders will be approximately 1,000 containers. This marks that the construction and upgrading of data centers has entered a new stage globally, and the application of artificial intelligence technology will continue to deepen.

Overall, as the demand for computing power surges, technological changes in AI servers are imperative, and water-cooling design will become an important direction for future development. The future of artificial intelligence will become increasingly bright and worthy of the expectations and attention of all walks of life.

FAQ

Who is the AI course suitable for?

AI courses are suitable for people who are interested in artificial intelligence technology, including but not limited to students, engineers, data scientists, developers, and professionals in AI technology.

How difficult is the AI course to learn?

The course content ranges from basic to advanced. Beginners can choose basic courses and gradually go into more complex algorithms and applications.

What foundations are needed to learn AI?

Learning AI requires a certain mathematical foundation (such as linear algebra, probability theory, calculus, etc.), as well as programming knowledge (Python is the most commonly used programming language).

What can I learn from the AI course?

You will learn the core concepts and technologies in the fields of natural language processing, computer vision, data analysis, and master the use of AI tools and frameworks for practical development.