Meta, formerly known as Facebook, has unveiled its in-house infrastructure for AI workloads, including generative AI, with the aim of keeping pace with rivals such as Google and Microsoft. The company has struggled to turn many of its more ambitious AI research innovations into products, particularly on the generative AI front. Meta has made plans to start developing a more ambitious in-house chip, due out in 2025, capable of both training AI models and running them. The new chip is called the Meta Training and Inference Accelerator (MTIA) and is part of a family of chips for accelerating AI training and inferencing workloads. Meta is also developing another chip called the Meta Scalable Video Processor (MSVP) for the processing needs of video on demand and live streaming. Meta’s Research SuperCluster (RSC) has completed its second-phase buildout and contains a total of 2,000 Nvidia DGX A100 systems sporting 16,000 Nvidia A100 GPUs. The RSC confers the benefit of allowing Meta’s researchers to train models using real-world examples from Meta’s production systems. Meta is feeling increasing pressure from investors concerned that the company’s not moving fast enough to capture the potentially large market for generative AI. The total addressable market for generative AI software could be $150 billion. Even a small slice of that could erase the billions Meta has lost in investments in metaverse technologies.
