Meta built their own AI chip. It's bigger than you think. #meta #MTIA #SemiAnalysis

Mark Zuckerberg is on a massive spending spree. Meta is basically buying everything they can and that includes GPUs from Nvidia and AMD and even TPUs from Google, which makes it easy to forget that Meta has its own silicon. The Meta Training and Inference Accelerator, MTIA for short, is now in its 3rd iteration. We don't know all the specs yet, but it will be produced in TSMC's N3P with very likely over 100 billion transistors and HBM memory. Meta knows exactly what their internal workloads need, and that's not only AI chatbots, but AI recommendation models that run the algorithm for Facebook, Instagram, and Threads. MTIA v3 won't be Meta's chip for AGI, but it offers good margins for their business model. And that means they can use all the external hardware to train new models, while internally they switch to their own silicon for inference. Looking at the roadmap, We will see a lot more MTIA in the future.