The alliance between AI chip hardware and software giants has attracted the attention of the market. Prior to the stock market on August 19th, American AI chip design giant AMD announced its acquisition of ultra large scale AI and cloud storage service provider ZT Systems.
The acquisition of ZT Systems marks an important step in our long-term artificial intelligence strategy, "said AMD CEO Zifeng Su. AMD is fiercely competing with strong rivals such as Nvidia in the AI chip market, and enhancing its software capabilities is crucial for boosting its server sales.
AMD's business in the field of artificial intelligence will continue to expand, including the development of training and inference chip hardware. However, in order to enable cloud service providers and enterprise customers to deploy these hardware quickly and on a large scale, AMD needs to leverage ZT Systems' software capabilities and mature ecosystem to improve support for these customers in building servers. Through this acquisition, AMD is seeking an end-to-end data center infrastructure development.
AMD expects that the market size of data center AI accelerators will reach the $400 billion mark by 2027.
Seeking end-to-end data center infrastructure
After the announcement of the acquisition, AMD held a conference call. Su Zifeng emphasized at the meeting, "We must have optimization capabilities and professional knowledge at the system rack and even data center level. The complexity of large-scale clusters is increasing day by day, which is precisely why we acquired ZT Systems
ZT Systems, headquartered in Secaucus, New Jersey, USA, has a history of over 29 years since its establishment in 1994. The vast majority of the company's customers are large-scale cloud service providers and top global hyperscale enterprises, capable of delivering hundreds of thousands of servers and tens of thousands of AI racks annually. With its profound expertise in data center systems and a well-established OEM and ODM ecosystem, ZT Systems has become a leader in the server market.
As a well-known chip design giant, AMD has multiple well-known artificial intelligence chips under its umbrella, including high-performance Instinct AI accelerator and EPYC (Xiaolong) CPU. After the transaction is completed, ZT Systems will be merged into AMD's Data Center Solutions Division.
According to market calculations, the addition of over 1000 engineers under ZT Systems is expected to increase AMD's annual operating costs by $150 million. However, this also means that AMD has introduced a high-quality design and customer support team, which is conducive to accelerating the deployment of the company's AI rack level systems in the cloud and enterprise level customers.
According to the acquisition arrangement, Frank Zhang, CEO of ZT Systems, will lead the manufacturing business of the new department, while Doug Huang, President, will be responsible for the design and customer support teams, both of whom will report to Forrest Norrod, Executive Vice President and General Manager of AMD.
However, AMD plans to divest ZT Systems' data center infrastructure manufacturing business in the United States after the acquisition is completed and seek strategic partners to acquire this part of the business. ZT Systems' manufacturing operations are spread across multiple regions worldwide, including the United States, Europe, the Middle East and Africa, as well as the Asia Pacific region.
The above acquisition arrangements reflect AMD's next step in AI strategy, which is a comprehensive improvement in chips, software, and systems.
AMD plans to spend $4.9 billion on the acquisition through cash and stock, including $3.675 billion in cash and $1.225 billion in AMD stock. If specific performance targets are achieved after the transaction, AMD will also pay an additional $400 million.
Enhance the competitiveness of AI chips
In the field of AI chips, AMD hopes to challenge Nvidia's dominant position. In 2024, Nvidia's data center revenue is expected to exceed $100 billion, while AMD's MI series accelerator chip revenue is expected to reach $4.5 billion. Although the gap in strength between the two is still significant, AMD has not given up on competition.
Based on recent financial data, the data center business is the main driving force behind the growth of these two GPU giants and will maintain high-speed growth in 2024.
Nvidia set a quarterly revenue record of $26 billion in the first quarter of fiscal year 2025 (February April 2024), with data center business contributing $22.6 billion and an astonishing annual growth rate of 427%. As a comparison, AMD's revenue for the first and second quarters of 2024 was $5.5 billion and $5.8 billion, respectively, with data center sales reaching $2.3 billion and $2.8 billion, representing year-on-year growth of 80% and 115%, respectively.
In terms of profits, Nvidia is also ahead of AMD. In the first quarter, AMD's core operating profit was only $415 million, a decrease of 45.1% compared to the previous quarter. In the similar period of February to April, Nvidia's net profit reached $14.88 billion, a year-on-year increase of 628%.
In terms of market value, Nvidia is currently leading the global chip industry with a market value of $3.2 trillion. As of the close of the US stock market on August 19th, AMD's market value was $251.32 billion, only about 8% of Nvidia's.
In 2024, the development of big models and generative AI has put forward new requirements for computing power, algorithm platforms, and data. Traditional computing infrastructure can no longer meet the needs, and a new AI platform service model is needed. The capital market's attention to the AI field is focused on infrastructure deployment and application development, and infrastructure deployment has led to a sustained growth in demand for hardware suppliers such as NVIDIA, AMD, and Intel.
According to an industry report by Mordor Intelligence, the size of the artificial intelligence infrastructure market is expected to reach $68.46 billion by 2024 and grow at a compound annual growth rate of 20.12% during the forecast period (2024-2029), reaching $171.21 billion by 2029. Among them, the market size of data center accelerators is expected to reach $11.29 billion by 2024 and $21.92 billion by 2029, with a compound annual growth rate of 14.19% during the forecast period (2024-2029).
In order to expand its influence in the AI chip industry, AMD has been taking frequent actions recently. In the past 12 months, AMD has invested over $1 billion to expand its AI ecosystem and enhance its AI software capabilities. Among them, AMD has completed the acquisition of Silo AI, the largest private laboratory in Europe, with a transaction value of approximately $665 million; AMD has also participated in a $1 billion funding round for AI data platform Scale AI and invested in artificial intelligence startup Firework AI, demonstrating AMD's ambitious goals in the AI field.