AMD confirms Xilinx merger approved by regulators

While Nvidia loses an Arm, this acquisition has legs


AMD on Tuesday said it has passed all the regulatory hurdles to complete its $35bn acquisition of Xilinx, which will close on Monday.

The acquisition of Xilinx will bulk up AMD's product offerings with FPGAs (that's field programmable gate arrays), which are reprogrammable chips used for all sorts of applications, from accelerating machine-learning software to prototyping chips and providing glue logic to handling network traffic at the edge.

The acquisition was announced in October 2020, but the closing was delayed as AMD waited for China to approve the deal, which happened late last month.

AMD’s completion of the multi-billion dollar acquisition is a milestone of sorts, with nations shutting down deals amid sanctions and security concerns surrounding semiconductors. Nvidia this week walked away from acquiring Arm because of government opposition, and China in 2018 shut down Qualcomm's $44bn acquisition of NXP.

AMD said it will combine its CPUs, GPUs, and FPGAs for cloud and edge offerings. On an earnings call last week, AMD CEO Lisa Su said the company's "customers are anxious to talk to us about combined road maps."

Xilinx will give AMD an extensive enterprise customer list, which will open up opportunities in markets such as communications and 5G infrastructure. Xilinx chips are also used in spacecraft, industrial systems, embedded electronics, and other markets.

"As we bring Xilinx into the equation … they also have very deep relationships with a number of these accounts. And so, you know, we see that as an incremental positive as we think about Epyc in communications," Su said.

AMD's Epyc server chips are being increasingly adopted in cloud and data center infrastructures, with Facebook being a notable customer. Offering FPGAs alongside CPUs and GPUs could be appealing to data center customers, AMD is thinking, and recent experience might well bear that out.

FPGAs are more like blank-slate chips that have reprogrammable circuitry; engineers describe the chips' desired operation using a high-level language, such as a hardware description language like System Verilog or VHDL, or even something more mainstream like C/C++ and others these days. This code is compiled down into a bitstream that is fed into the FPGA to configure its internal logic and signal pathways.

That means the gate arrays can be customized to handle specific tasks fast in hardware, and they today typically ship with baked-in units, such as network and other IO interfaces, dedicated CPU cores, cryptography engines, and more. For instance, FPGAs could be configured to handle network filtering to free up their host servers' main processors and chipsets.

Amazon and Microsoft, to name two, offer FPGAs in the cloud.

AMD’s biggest FPGA competition will be its x86 rival, Intel, which acquired FPGA biz Altera for $16.7bn in 2015. Smaller rival Lattice Semiconductor meanwhile just recently acquired a software outfit to aid the acceleration of AI operations on its hardware. ®

Similar topics

Narrower topics


Other stories you might like

Biting the hand that feeds IT © 1998–2022