Facebook Inc. (NASDAQ:FB) is planning to step into a whole different realm altogether. The idea to develop its own chipset by a company, known for its dominance over the internet, is a bit odd, albeit Facebook seems to be working on exactly the same. The Social media giant may join other tech moguls that are lowering the dependency on chipmakers such as Qualcomm and Intel Corp.
Facebook is forming a team to design its own semiconductors, though the work is still in early stages, according to a new report from Bloomberg. Facebook has posted a job opening seeking a design expert in ASIC and FPGA, two custom silicon designs, or say a unique type of ICs, that can be utilized in specific use cases – particularly in Machine Learning and AI. The company is looking to hire a manager to create an “end-to-end SoC/ASIC, firmware and driver development organization”. Facebook AI researcher tweeted about the role and pointed to the job listing.
The FAMGA group has already embarked on the in-house hardware making road and apparently Facebook is last but not the least. Google produces custom AI chips and included its first consumer-focused silicon in Pixel 2 last year. Amazon is making custom hardware to improve its AI Alexa and connected devices. And recently, Apple is planning to make its own chip for Mac, thereby replacing Intel. Now, Facebook too is gearing up for customized hardware domain. The world’s largest social network is gradually becoming more self-reliant and less dependent on chipmakers.
Just a FYI: A system on chip, or SoC, is a silicon-based processor typically used in mobile devices. ASIC, or application specific integrated circuit, is a chip designed for specific purpose. FPGA, or Field Programmable Gate Array, is also a customized piece of silicon which is known to be more flexible and able to adapt to changing the ML-driven landscape. Both ASIC and FPGA are also used in particular works, like cryptocurrency mining.
Facebook was not into devising customized hardware range yet. The company is currently working on a slew of smart speakers. Next month, it is planning to launch a standalone VR headset Oculus Go, a $200 all-in-one VR gear that runs on a Qualcomm processor. By using its own processor chips, Facebook can gain an extra edge in the hardware segment. the company could deploy chips to power its hardware devices, AI software and servers. This would give it a better control over product development.
Nvidia has a huge control over the AI space with its GPU technology, which it can optimize for frameworks like TensorFlow, an open source machine learning framework. Even the servers that train Facebook’s AI system are Nvidia-powered. Mark Zuckerberg has said that AI will play a bigger role in flagging inappropriate content across the social media platform.
Though the job posting has confirmed Facebook’s custom hardware aspirations, its still unclear that what and how is Facebook going to use the chips, besides signalling a broad umbrella of AI. Of course, this idea is not worth calling amazeballs but seems shady considering the very recent privacy abuse and misuse of user data. The obstinate will think that does making chips or enhancing AI would make them safer online or it might make Facebook spy on them more easily.
Is it a viable move, considering Facebook’s deeper issues that continue to mount?
If only it helps in protecting user privacy, even a little, it sure is.