Meta (formerly Facebook) has been branching out of late into fields that are not as social media focused. To that end the company recently detailed its new AI supercomputer – the AI Research SuperCluster (RSC) as it terms it.

The company is currently in the process of fully building out this supercomputer and once is it done, which is estimated to be mid-2022, Meta says it will be the fastest supercomputer of its kind in the world.

This as it should be capable of, “performing at nearly 5 exaflops of mixed precision compute.” Powering this performance is 760 Nvidia DGX A100 systems, with a total of 6 080 GPUs.

From its early benchmarking, Meta says the RSC system can run computer vision workflows up to 20 times faster and the NVIDIA Collective Communication Library more than nine times faster than the previous platform the company had in place.

So what is the endgame here, apart from laying claim to having the world’s fastest AI supercomputer? If you guessed the metaverse, you guessed right.

“RSC will help Meta’s AI researchers build better AI models that can learn from trillions of examples; work across hundreds of different languages; seamlessly analyze text, images and video together; develop new augmented reality tools and more,” explains Meta in a blog post.

“Ultimately, the work done with RSC will pave the way toward building technologies for the next major computing platform — the metaverse, where AI-driven applications and products will play an important role,” it adds.

With several other big tech firms also looking to lay claim to the metaverse, it will be interesting to see whether they too opt for an AI supercomputer platform to get them there, like Meta is currently doing.

That said, we don’t think Meta will be sharing access to the RSC anytime soon, so they will need to look elsewhere to find the necessary AI compute.