:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/63709554/altera_chip_wiki_commons.0.1536949917.0.jpg)
If published reports that chipmaker Intel intends to acquire another chip company, Altera, turn out to be true, it would likely kick off a rush by other tech companies to acquire Altera’s primary rival, Xilinx.
Shares of Altera closed up more than 28 percent after the Wall Street Journal reported today that Intel is in talks to acquire Altera for a price said to be above $10 billion. Shares of Xilinx, which competes with Altera in the market for a type of chip known as a Field Programmable Gate Array, rose by more than 6 percent.
A deal of that size would be Intel’s largest takeover ever and would add about $2 billion in annual revenue. Intel and Altera are already closely tied. Intel not only manufactures Altera’s FPGA chips under contract — Altera doesn’t own its own chip factories — but also offers some of its Xeon chips for servers combined with Altera’s FPGA chips.
Analyst Pat Moorhead, head of the research firm Moor Insights and Strategy, says FPGA chips are generally used for two things. When designing their products, chip companies often use FPGA chips to create early prototypes to test ideas before those ideas are locked into a final silicon design, their main advantage being their flexibility. They can be tweaked and reprogrammed to try different ideas, he said.
That flexibility also makes FPGA chips like those created by Altera and Xilinx useful in speeding up certain specialized computing work loads. “Because they’re flexible and programmable, you can tune an FPGA chip to do a certain kind of computing job really fast, and then offload that job from the main CPU chip like an Intel Xeon,” Moorhead said. That makes them highly useful in certain demanding environments in data centers. Microsoft has used them to speed up systems powering its Bing search engine, and Google is thought to do the same, though it has never confirmed it.
If Intel reaches a deal to acquire Altera, then it follows that speculation will ramp up about an acquisition of Xilinx. Among the potential buyers are IBM and Qualcomm, Moorhead says.
While IBM no longer owns a chip manufacturing unit, it still has some skin in the chip game. It owns the designs for its Power chips inside its specialized servers. And last year it released the designs of that line as an open-source project, meaning they can be manufactured by anyone with the means to build them.
And like Intel’s Xeon chips, OpenPower chips can work in combination with an FPGA chip, including those produced by Altera and Xilinx.
“One of the linchpins of IBM’s strategy with OpenPower is that it works with an FPGA chip, and if one of the FPGA players is about to get gobbled up, that means the other might soon follow,” Moorhead said.
Xilinx is worth about $11 billion as of close of trading today, meaning it could be acquired for about $13 billion, give or take. IBM, which has about $8 billion in cash on the balance sheet, would have to take on debt in such a deal, and it would be a big bite.
Another potential buyer is Qualcomm, which last year said it would get into the business of designing server chips based on the ARM architecture. Owning an FPGA chip producer could bolster that business, Moorhead said, and Qualcomm, which has about $18 billion in cash and short-term investments, could potentially afford Xilinx. But even so, Moorhead put Qualcomm lower on the list of likely buyers.
This article originally appeared on Recode.net.
