Produced by 21ic China Electronics Network Fu Bin
Website: 21ic.com
In the context of accelerating heterogeneous computing, “giant mergers” have become synonymous with this year.
On October 9, The Wall Street Journal reported that American processor company AMD (Advanced Micro Devices) is in deep negotiations to acquire its competitor chip manufacturer Xilinx, with a deal potentially exceeding $30 billion.
Insiders claim that the transaction agreement could be finalized as early as next week, but there is no guarantee that the deal will be completed. The source emphasized that the acquisition had previously stalled, making the outcome unpredictable; however, both parties have resumed discussions, accelerating the transaction process.
Xilinx, when contacted by 21ic China Electronics Network, stated that it would not comment on rumors or speculation.
In terms of market capitalization, AMD currently exceeds $100 billion, while Xilinx’s market value approaches $26 billion.
The Golden Partnership of the Heterogeneous Computing Era
“AMD Yes” has become the most significant praise from netizens regarding AMD’s gradual rise to prominence. Since Dr. Lisa Su became president and CEO in October 2014, her strong yet approachable style has earned her the affectionate nickname “Mama Su” among fans. Particularly, the Ryzen and EPYC processors have achieved remarkable success from laptops to desktops and data centers. In the graphics card sector, AMD has also fiercely competed with NVIDIA, winning the favor of Sony, Microsoft for their consoles, and Samsung for their smartphones.
Xilinx, primarily focused on FPGAs (Field Programmable Gate Arrays), has a strategy centered on three main areas: “data center first,” “accelerating core market development,” and “driving adaptive computing.” This year, it has launched the integrated SmartNIC platform Alveo U25, the most powerful 7nm cloud chip Versal Premium, and an innovative TCON (Timing Controller) solution for FPGA devices.
As semiconductor development progresses, an unavoidable fact is that Moore’s Law is slowing down. With the slowdown of Moore’s Law, the Dennard scaling law and Amdahl’s Law nearing their limits, Moore himself once proposed a “cure”—heterogeneous computing. We are now in the “golden age” of heterogeneous CPUs and accelerators.
In fact, a similar scenario played out back in 2015 when Intel acquired FPGA manufacturer Altera for $16.7 billion, which subsequently provided a solid foundation for Intel’s “CPU + xPU (GPU + FPGA + ASIC + eASIC)” strategy.
AMD and Xilinx have maintained a close collaboration, previously providing a series of storage system IPs such as NVMe HA, NVMe TC, and Embedded RDMA for AMD’s EPYC (EPYC) data center processors, which help AMD build low-latency, efficient data pathways, thus enabling efficient FPGA storage acceleration.
21ic believes that the accelerated computing of CPU + GPU + FPGA undoubtedly targets the blue ocean of the data center field. Intel has repeatedly stated that it is a data-centric company, while NVIDIA has recently shown its determination to “occupy the high ground” in its acquisition proposals and new product releases…
This year may see three major semiconductor acquisition deals.
After several acquisition shocks, it seems that merger news is no longer surprising.
In July 2020, American chip giant Analog Devices Inc. (ADI) announced plans to acquire competitor Maxim Integrated Products for $20.9 billion in an all-stock deal to enhance its capabilities across multiple industries, including telecommunications. This was the largest merger deal in the U.S. at the time and ADI’s largest acquisition ever.
In September 2020, NVIDIA and SoftBank Group Corp. (SBG) announced that they had reached a definitive agreement. Under the agreement, NVIDIA would acquire Arm Limited from SBG and the SoftBank Vision Fund for $40 billion. This transaction is expected to immediately increase NVIDIA’s non-GAAP gross margin and non-GAAP earnings per share.
In October 2020, American processor company AMD was in deep negotiations to acquire competitor chip manufacturer Xilinx, with a deal potentially exceeding $30 billion.
Notably, just last week at GTC, NVIDIA announced the launch of a new type of processor—DPU (Data Processing Unit), supported by a new DOCA (Data-Center-Infrastructure-On-A-Chip Architecture) architecture—a completely new data center IOC (Infrastructure On A Chip) architecture that enables breakthrough network, storage, and security performance.
Its solutions include multi-core CPUs based on Arm architecture, GPUs for processing AI applications, and high-speed network transmission interfaces, the latter coming from the SmartNIC technology acquired from Mellanox. After receiving regulatory approval, NVIDIA announced the completion of the Mellanox acquisition in April this year.
Even Huang Renxun clearly stated: “The data center has become a new type of computing unit. In modern, secure, accelerated data centers, DPUs have become an essential component. The combination of CPU, GPU, and DPU can form a fully programmable single AI computing unit, providing unprecedented security and computing power.”
Intel, on the other hand, announced at its 2020 Architecture Day the combination of CPU + discrete GPU + FPGA + AI accelerators and the competitive “10nm SuperFin technology” for process node transitions. A series of signs undoubtedly lead the industry to speculate that this acquisition is AMD’s rapid response to its competitors, driven by a series of merger transactions.
21ic believes that the golden partnership of CPU and FPGA in the heterogeneous era is undoubtedly the best combination for accelerating data centers. With Intel and NVIDIA’s recent major news, the combination of AMD and Xilinx is a comprehensive enhancement of the ecosystem under the “pressure” of competitors. Of course, Xilinx itself possesses strengths in FPGA, storage system IP, SmartNIC, etc., making it fully capable of handling the acceleration computing tasks of AMD’s CPUs and GPUs.
Related News Reading
【1】Xilinx’s Most Powerful 7nm Chip Competes in the Cloud Market【2】Under Three Strategies, Xilinx Depicts the Trinity of Data【3】Embracing Open Source, Xilinx Depicts a Balance of Strength and Flexibility Under AI
Valuable Skills Articles
【1】This Article on PCB Reflow is All You Need!
【2】A Must-Read for PCB Practitioners: Special Routing Techniques and Tips!
Valuable! Protection Circuits in Port Design
What is the Difference Between You and Top Engineers?Join the Technical Exchange Group for Face-to-Face Interaction with ExpertsAdd the Administrator’s WeChat