Daily Management Review

Facebook may start production of its own microprocessors


04/19/2018


Facebook social network can start producing its own microprocessors for use in mobile devices. This was reported by the American media, referring to Facebook job ads that are looking for experts to develop the appropriate chips.



pexels
pexels
Media assumptions are based on Facebook’s announcement about hiring specialists in the development and production of ASICs and FPGAs. The announcement says that the required specialists will work in the FB infrastructure team and will interact with system and software engineers of FB "to understand current limitations of hardware equipment and use their experience in building client solutions related to data compression, video coding, etc.".

The American media assume that such an expert in development of hardware can easily begin to create a team whose goal will be to devise their own microprocessors aimed at solving specific problems relevant to the social network: processing large data sets, artificial intelligence (AI) servers, etc.

Thus, Mark Zuckerberg can solve two problems at once: to make the product to solve their own tasks and not depend on external suppliers, for example Intel and Qualcomm.

Bloomberg recalls that Google and Apple similar way already went to companies such as. In addition, there are suggestions that Facebook can use its own experience in the production of their mobile devices, such as helmets Oculus. As you know, in March 2014 Facebook bought the manufacturer of these helmets company Oculus VR, for $ 2.3 billion. 

Techcrunch, citing sources in the industry, notes that FB is currently actively working on developing its artificial intelligence systems and processing user data. The publication, in particular, mentions an AI infrastructure called Caffe2, used by FB, and that the social network has long wanted to optimize for this infrastructure hardware equipment and, possibly, microprocessors. In turn, FPGA modules are responsible for more accurate tuning of machine learning systems and allow to solve more related problems.

source: techcrunch.com