Xilinx deal is a push for cloud computing
This week, AMD announced that it has reached an agreement to acquire semiconductor maker Xilinx for $35 billion. The proposed deal is intended to transform AMD, giving the company more resources and new business lines to better compete with Intel, especially with chips designed for data centres, networking and other cloud computing applications. AMD’s recent growth, ballooning share price and $92 billion market capitalization has provided the basis for the deal, with AMD using equity rather than cash for the transaction.
Much of this growth has come from AMD’s success in the PC industry with CPUs and GPUs. The combination of a much-improved product line-up under CEO Lisa Su and Intel’s manufacturing misfires have helped AMD win back market share.
The Xilinx transaction sees AMD setting its sights on Intel’s data centre business and growth beyond its marginal position in the segment today. Intel dominates the market for data centre CPUs thanks to its Xeon silicon and investment in alternative architectures such as field-programmable gate arrays (FPGAs). In the third quarter of 2020, Intel’s data centre unit accounted for $5.9 billion of the company’s $18.3 billion of revenue (see Instant Insight: Intel Results, 3Q20).
AMD will be hoping to get a bigger slice of a lucrative and high-margin market, and further benefit from its use of leading-edge process technology at a time when Intel’s manufacturing strategy is facing significant questions. Part of the attraction of Xilinx is that it offers much higher gross profit margins and free cash flow generation.
The Xilinx acquisition will add several new capabilities to AMD’s portfolio, including FGPAs, which, unlike a more general-purpose CPU or custom-built application-specific integrated circuit, can be reprogrammed after manufacture or as requirements change once deployed.
FPGAs are useful with emerging technology such as 5G infrastructure, which doesn’t have many existing microprocessors designed specifically for the purpose. With FPGAs also gaining traction in wireless networks, the deal opens the door to new telecom customers for AMD, just as the industry spends billions of dollars to build 5G networks. FPGAs are being used in many new bleeding-edge applications in machine learning, communications equipment and many other devices. This also extends to automotive applications and aerospace, although Xilinx has been hurt by geopolitical tensions between China and the US. Another major FPGA supplier is Intel, which built out its own FPGA capability by buying Altera in 2015.
The US semiconductor industry is going through a huge shift, sparked by a wave of corporate consolidation, growing demand for cloud computing, the rise of artificial intelligence and a pandemic that has increased hunger for processors. In September 2020, graphics chip-giant Nvidia agreed to pay $40 billion for Arm in what could be the industry’s biggest deal if completed — but that’s a big if. This also stands to put more pressure on Intel’s server business, with Nvidia keen to use Arm to develop server CPUs that can be more tightly coupled with its GPUs (see Instant Insight: Nvidia Acquires Arm). And last week, Intel agreed to divest its NAND memory business to SK hynix for $9 billion.
Intel and AMD, bitter rivals for decades, have become household names to people familiar with the innards of PCs. With its latest acquisition, AMD will take the fight to Intel in the data centre and networking markets. Intel’s pedigree and heritage will make this a challenging task, but its manufacturing troubles coupled with the speed of AMD’s rise in PCs are a reminder that the new AMD is a force to be reckoned with.