COMPUTE Contemplating Meta’s Homegrown MTIA Compute Engine Roadmap It is very hard to guess what an AI compute engine should look like five years into the future, other than the obvious more, more, more. …
… Rene Haas, the chief executive of Arm shown above holding the first generation AGI CPU that is sampling now and that will ship in volume later this year to Meta Platforms and OpenAI and anyone else who wants to place an order, laid out the math for why CPUs still matter in agentic AI datacenters. …
… Nvidia Mostly Owns Training, And Can Compete On Inference These roadmaps are important to the OEMs and ODMs that convert Nvidia’s technology into the systems that run AI training and inference for the vast majority of the world. …
… There is every reason to believe that Meta Platforms can build up its AI platform cloud business fairly quickly, and it all starts from the groundswell of support for an open source LLM that could compete with the best of the proprietary models like those from OpenAI, Google, and Anthropic. …
… The Forthcoming IPOs It also gives Microsoft more running room to develop and run more of its own AI models, a time when rivals like Meta Platforms – with open source Llama and closed source Muse Spark – and Google via Gemini are aggressively pushing ahead with their AI roadmaps. …
… If the spending is doubling, the compute capacity of the gear is also probably doubling to quadrupling, depending on the computing precision used, as AWS moves through the Nvidia roadmap and its own Trainium roadmap. …
… Tan added that in contrast to what many rumors say about the MTIA accelerator effort at Meta Platforms, “Meta’s custom accelerator MTIA roadmap is alive and well” and it is shipping now. …
… Here is how SambaNova models the SN50 running the InferenceX benchmark from SemiAnalysis using the Llama 3.3 70B model from Meta Platforms: That SN50 curve is shaped kinda funny, isn’t it? But it has a lot higher throughput on inference, and maintains it to a over a longer speed per user range. …