The chips, code-named Lake Crest, would then begin sampling to key customers later that year.
It was one of a few product announcements Intel made at an AI-themed event Intel held in San Francisco on Thursday. Speakers included CEO Brian Krzanich and Diane Bryant, executive vice president of Intel’s data center group — indications of how seriously Intel is about artificial intelligence.
Lake Crest will be followed by Knight’s Crest, which will integrate Nervana’s architecture with Intel’s Xeon microprocessors.
The roadmap for Xeon Phi, Intel’s own line of machine-learning processors, will continue as well. The current Xeon Phi generation is called Knight’s Landing, and it will be followed by Knight’s Mill, due out in 2017.
In terms of processors, machine learning and deep learning have been more the purview of Nvidia and its GPUs. The acquisition of 2-year-old Nervana, announced in August, is meant to give Intel some ammo in that fight, as Nervana had developed technology for deep-learning chips.
Intel also has an ally in Google. The companies announced today that they’ll work together to advance AI in multicloud environments. Kubernetes and TensorFlow, an open source AI library, will figure in that work.
On a more conventional front, Intel announced that the next generation of Xeon processors, Skylake, has begun shipping, but only to certain cloud providers. That probably means Facebook and Google, analyst Amit Daryanani of RBC Capital Markets writes in a research note issued today.
Daryanani also points out that Intel doesn’t normally focus its early samples of chips this way; the company usually samples to a broader market.