Amazon, the world’s largest online retailer and largest cloud-computing company, is pushing into a new line of work: computer chips.
Late last month, the company, based in Seattle, revealed that it had spent the last few years building a new chip for use inside the millions of servers in its data centers around the world.
Amazon does not plan to sell this chip directly to customers, but the decision by one of the world’s biggest buyers of computer processors to go the do-it-yourself route is likely to have a major impact on Intel, the iconic Silicon Valley chip maker.
Intel has struggled to keep up with recent technology trends, from the increasing use of mobile devices to artificial intelligence, as influence over chip designs has been shifting to Amazon, Apple and Google, which already dominate much of the rest of the tech industry.
Amazon’s new chip arrives as market forces are rapidly undercutting chip makers and their $412 billion in yearly sales. Online operations like Amazon and Google have grown so large, they can save significant money by making chips tailored to their needs rather than buying them from longtime suppliers.
The homegrown chip also gives Amazon bargaining power with Intel, which hasn’t had much competition in the server market in recent years. The internet company will still buy from Intel because it would be difficult to build all the chips it needed on its own. But now it will have options.
Intel acknowledges that companies like Amazon want to reduce their dependency on one big chip supplier. “I am not unaware of the dynamic,” Lisa Spelman, an Intel vice president, said. “Our goal is to understand and respectfully acknowledge that desire.”
About 35 percent of the server chips sold around the world go to about 10 companies, including large internet companies like Amazon and a handful of telecommunication firms, said Shane Rau, a chip analyst with the research firm IDC. That just one of them is shifting its business is terrible news for Intel.
“Each one of these companies is so large, they represent a market unto themselves,” Mr. Rau said.
In recent years, Google has designed specialized chips for artificial intelligence technology. Facebook and Microsoft, which like most internet companies are major buyers of chips from Intel, have indicated that they are working on similar A.I. chips.
Apple beat the other tech giants to this cost-saving trend four years ago when it unveiled its first custom-built chip for the iPhone. Google and Microsoft are also building the chips that go into devices like smartphones and virtual-reality headsets.
Amazon has upped the ante. In 2015, it spent a reported $350 million to acquire a chip maker, Annapurna Labs, which helped build the new central processing unit, or C.P.U.
A C.P.U. — the centerpiece of a computer’s operations — is the sort of chip that Intel has made for decades to run personal computers and servers. Building these chips requires rare expertise and hundreds of millions of dollars in capital. It is a big step up in complexity from building chips tailored to certain tasks.
“The belief was that you needed some magic to build a processor, particularly a server processor,” said Andrew Feldman, the chief executive of the chip start-up Cerebras and a former executive at the chip maker AMD. “You had to be Intel or AMD — and that was about it.”
Amazon executives believe the chip, which was designed to be more energy efficient, will help reduce the cost of electrical power in its data centers. It said it was offering a cloud-computing service that would allow business customers to use its new chip. The cost of the service could be 45 percent lower than other options.
And when Amazon buys chips from other companies, a homegrown option gives it even more sway over prices. “They can now say to Intel, ‘We will just move the workloads to other chips,’” Mr. Feldman said.
Amazon has also designed a chip for artificial intelligence, called the Inferentia. In a recent blog post, James Hamilton, vice president and distinguished engineer inside Amazon’s cloud computing arm, said the company would continue building new chips for artificial intelligence and other specialized tasks.
