Microsoft has quietly installed an esoteric, powerful piece of silicon across its cloud-oriented data centers, and it’s saying the new chips will give its Azure cloud computing platform a strong competitive advantage in the burgeoning field of artificial intelligence.
Using the technology, the company demonstrated what it called “the first AI supercomputer” during Microsoft CEO Satya Nadella’s closing keynote address on the first day of the company’s Ignite technology conference in Atlanta.
It’s part of Microsoft’s vision to “democratize” artificial intelligence, Nadella said. He pointed to four pillars: agents, applications, services, and infrastructure, with the new chips being an example of the latter.
Field programmable gate arrays, or FPGAs, have been quietly installed in Microsoft servers in 15 countries on five continents over the past two years, said Doug Burger, Microsoft Research’s principal hardware-development engineer on what’s known as “Project Catapult.”
The initiative’s roots inside Microsoft date back more than six years, according to a lengthy story in Wired detailing the project.
FPGAs are similar to the central processing units (CPUs) and graphical processing units (GPUs) that make up every computer’s logic fundamentals. But they can be reprogrammed on the fly to best serve whatever computing task is at hand.
“What that means is you get the efficiency of hardware but you also get flexibility because you can change their functionality on the fly,” Burger said. “This new architecture that we’ve built effectively (put) an FPGA-based AI supercomputer into our global, hyperscale cloud. We get awesome speed, scale and efficiency. It will change what’s possible for AI.”
As an example of what’s possible, Burger showed the ability to translate 5 billion words into another language into less than one-tenth of a second — half the amount of time it takes for a human to blink. “That crazy speed shows the raw power of what we’ve deployed in our intelligent cloud.”
Burger said Microsoft is “the first to have its global hyperscale cloud enhanced with FPGA’s. It gives us the most powerful, the most flexible and the most intelligent cloud.” He said FPGAs confer “ten times the AI capability of the largest existing supercomputer” so that “we can solve problems with AI that weren’t even possible before.”