Stephen Nellis
SAN FRANCISCO, Dec 2 (Reuters) – Silicon Valley-based startup Axiado, which is making a chip designed to save space and power in artificial intelligence servers, said on Tuesday it had raised $100 million in new funding.
Within an AI data center, thousands of powerful servers must work together to train models. Doing so requires coordination over an internal network, and all servers contain a small number of chips whose job is to process these instructions and manage the server’s circuit boards while ensuring they are safe from hackers.
These motherboard management chips tend to use older technology, and Axiado is working to consolidate all of them into a smaller chip. Currently, major chip companies such as Nvidia and Advanced Micro Devices are trying to pack as many graphics processing units as possible into servers, while data centers are trying to make room for liquid cooling equipment to save space.
“There’s real value in the real estate out there,” Andrew Homan, managing partner at Maverick Silicon, which led the round, told Reuters. “If you can develop that footprint, that’s a huge value-add to any large enterprise that’s building these systems.”
But Axiado’s chips themselves also use artificial intelligence. When a given software program runs on a server, the chip can learn about the server’s normal behavior.
Using this knowledge, Axiado chips can check for any deviations from normal behavior that could represent a cybersecurity threat. Axiado’s chip can also learn to control the server’s cooling system, turning it up or down to save up to 50% of cooling energy.
“Our AI engine learns that behavior. Based on the behavior itself, I can tell that a particular load is going to take up a certain level of GPU capacity, so if you have full scale, you don’t need to run it,” Axiado founder and CEO Gopi Sirineni told Reuters.
(Reporting by Stephen Nellis in San Francisco; Editing by Lincoln Feast.)
