Semiconductor designer AMD said on Tuesday it would buy Xilinx in a US$35-billion all-stock deal, intensifying its battle with Intel in the data centre chip market.
The deal, which AMD expects to close at the end of 2021, will create a combined company with 13 000 engineers and a completely outsourced manufacturing strategy that relies heavily on Taiwan Semiconductor Manufacturing Co (TSMC).
The two US firms have benefited from a more nimble approach to grab market share from Intel, which has struggled with internal manufacturing. AMD has long been Intel’s chief rival for CPUs in PCs.
Since CEO Lisa Su took over at AMD in 2014, she has focused on challenging Intel in the fast-growing business of data centres that power Internet-based applications and services and are fuelling the rise of artificial intelligence and 5G networks.
Xilinx has also been working to penetrate data centres with programmable processors that help speed up specialised tasks such as compressing videos or providing digital encryption. Its primary rival in the area, Altera, was scooped up by Intel for $16.7-billion in 2015 in what was then Intel’s largest-ever deal.
Intel falls behind
“There are some areas where we’re very strong, and we will be able to accelerate some of the adoption of the Xilinx product family,” Su said in an interview. “And there are some areas where (Xilinx CEO) Victor (Peng) is very strong, and we believe that we’ll be able to accelerate some of the AMD products into those markets.”
The tie-up comes at a time when Intel’s manufacturing technology has fallen years behind TSMC’s. AMD, which spun off its factories nearly a decade ago, has rocketed ahead of Intel with chips that perform better. The performance edge helped AMD gain its best market share since 2013 at slightly less than 20% of the CPU market, which has in turn pushed its shares up 79% this year.
Shares of AMD fell about 4% in pre-market trading, while those of Xilinx surged over 12% after the deal was announced.
Xilinx also uses TSMC’s factories, called “fabs” in the industry, to make its chips, with both US companies using modular designs that let them swap out different pieces of a chip to avoid bottlenecks or delays.
“We ended up with TSMC, and have stayed with them, not due to any contractual reason — we could go to any fab at any time — but because they are best-in-class,” Peng said in an interview. “It’s about the choices you make.”
Under the deal, Xilinx shareholders will receive about 1.7 shares of AMD common stock for each share of Xilinx common stock, valuing Xilinx at $143/share, or about 24.8% higher than its $114.55 closing price on 26 October. AMD shareholders will own about 74% of the combined firm, with Xilinx shareholders owning the remaining 26%.
AMD’s Su will lead the combined company as chief executive, with Xilinx’s Peng serving as president responsible for the Xilinx business and strategic growth initiatives. The companies expect the deal to generate $300-million in cost savings.
Bernstein analyst Stacy Rasgon said there is a danger that a major acquisition in an adjacent chip market could distract AMD’s leadership while Intel fights to regain market share. “The worry would be, AMD has this great self-grown story of their own, which is just starting to play out. Why are you doing this now? Is it just opportunistic? Does it distract from the current story?” he said.
Xilinx’s Peng, however, said meetings between the two companies have already revealed they have very similar methods for designing chips. “I’ll be honest, I don’t think it’s really as challenging as some other combinations,” he said. “I had one of my leadership teams who was not familiar with AMD say to me after a meeting, ‘Boy, they’re just like us.'” — Reported by Stephen Nellis and Munsif Vengattil, (c) 2020 Reuters
