Intel has decided to bolster its artificial intelligence projects and efforts through acquiring Nervana Systems, a two-year-old startup considered to be one of the leaders in developing machine learning technology. Terms of the deal weren’t disclosed as of the time of writing, but a source familiar with the matter told VentureBeat that Intel paid more than $350 million.
Intel fleshing out AI portfolio
“Nervana’s Engine and silicon expertise will advance Intel’s AI portfolio and enhance the deep learning performance and TCO of our Intel Xeon and Intel Xeon Phi processors,” Diane Bryant, executive vice president and general manager of Intel’s Sata Center Group, said in a blog post.
While Intel has been running a successful business in high-performance computing (HPC), it has as of late taken a back seat to Nvidia, another HPC supplier, when it comes to creating chipsets for “deep learning.” A trendy type of of artificial intelligence, deep learning involves training artificial neural networks on lots of data and then getting them to make inferences on that new data.
Google has recently unveiled competing chips called “tensor processing units,” or TPUs, that can handle Google’s TensorFlow open source deep learning framework. At this point, it is uncertain if Google has teamed up with Intel in this regard. In any case, Intel’s acquisition of Nervana Systems represents a clear signal that Intel in invested in gaining traction in the artificial intelligence community.
Nervana to drastically improve Intel’s AI capabilities
“[Nervana’s] IP and expertise in accelerating deep learning algorithms will expand Intel’s capabilities in the field of AI,” said Bryant. “We will apply Nervana’s software expertise to further optimize the Intel Math Kernal Library and its integration into industry standard frameworks.”
Nervana Systems was founded in 2014 and is currently based in San Diego, California. The startup had raised nearly $25 million from investors including Data Collective, DFJ, Fuel Capital, Allen & Co, and Lux Capital. The company had captured attention early on with its hardware-centric approach to AI solutions and has since pursued technologies aimed at training neural nets.
“Intel didn’t just buy something that is simultaneously faster and more power-efficient that Nvidia,” said Data Collective managing partner Matt Ocko. “Intel bought something that it can sell on boards and systems and even supercomputers to its customers ready-to-go that out-punches anything that Facebook or Google or Baidu or Microsoft has.”