Intel has announced the acquisition of Habana Labs, with the chip giant shelling out $2 billion for the programmable chipmaking start-up based in Israel.
Habana builds programmable deep learning accelerators for data centres, and Intel is hoping the deal will be a big plus for its AI portfolio, as it looks to gain a bigger foothold in the AI silicon market.
Habana’s work revolves mostly around two main AI and machine learning products: the Gaudi AI Training Processor, and the Goya AI Inference processor. Gaudi is being sampled by a few hyperscale clients at the moment, and the media are saying it should deliver up to four times more throughput in large node training systems.
On the other hand Goya, already available for purchase, is designed to boost inference performance (both throughput and real-time lag).
Habana will keep its entire management team, but will report to Intel in the future.
"This acquisition advances our AI strategy, which is to provide customers with solutions to fit every performance need – from the intelligent edge to the data centre," said Navin Shenoy, EVP and GM of Intel's Data Platforms Group.
"More specifically, Habana turbo-charges our AI offerings for the data centre with a high-performance training processor family and a standards-based programming environment to address evolving AI workloads."