Google creates a vacuum of excitement wherever it goes. Since last year, when the company said that it had devised a custom chip for accelerating machine learning, there has been much chatter about the rise of specialized hardware for complex tasks like image recognition.
But the search engine company recently hinted that machine learning was not the only task that required custom silicon. In a recent overview of its data center security, Google said that it had devised “a hardware security chip” to verify the identity of devices inside its servers.
The chip is “currently being deployed on both servers and peripherals,” the company said. “These chips allow us to securely identify and authenticate legitimate Google devices at the hardware level.” It is unclear when the company first installed the chips, but they appear to be its latest strategy for storing security code.
Google added that with “each new generation of hardware, we strive to continually improve security: for example, depending on the generation of server design, we root the trust of the boot chain in either a lockable firmware chip, a microcontroller running Google-written security code, or the above mentioned Google-designed security chip.”
The existence of the security chip was first reported Monday by The Register, an information technology news website.
It is the latest sign that internet companies are increasingly willing to invest in custom chips that fit the unique demands of artificial intelligence. Microsoft is using in its servers custom FGPAs that can be programmed and optimized for an array of tasks. Facebook, too, has devised its own type of graphics processors.
Google’s chip creates a hardware root of trust, which entails locking a cryptographic signature in a special part of the chip where it cannot be altered. Google said that its machines validate that immutable code during each boot and update. The company did not immediately respond to a request for more information.
Read the source article at Electronic Design.