The number of neoclouds, which rent out AI computing power, has exploded over the past four years. Wall Street is watching closely because their business model, which moves tens of billions of dollars, raises questions.
HSBC's chief executive cautioned that not every data center, the campuses housing the servers behind AI and cloud computing, will be a sound investment despite the AI boom.
IBM and other contributors released version 1.0 of llm-d, an open-source distributed inference framework, ahead of PyTorch Conference North America. The release adds operational controls to help enterprises run large models reliably on their own infrastructure.
To meet demand for the memory chips packaged alongside AI GPUs, Samsung is building a huge manufacturing complex in Pyeongtaek, South Korea. Three fabs already run around the clock and a fourth is coming online.
In an op-ed in Le Monde, groups fighting data center projects warn of the water and electricity use and the loss of farmland these sites bring. They frame local opposition as a stand against the broader AI build-out.
DeepSeek released DeepInfer under Apache 2.0, the framework it uses to serve DeepSeek-V4.1-Flash in production. It targets NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs and currently supports offline inference only.