The race to build the world’s most powerful AI factories demands networking that keeps pace with the ambitions of AI itself. NVIDIA Spectrum-X Ethernet scale-out infrastructure stands at the forefront of that race as the most advanced AI networking technology available today, deployed by industry leaders who can’t afford to compromise on performance, resilience or […]
NVlabs releases cuda-oxide v0.1.0, a custom rustc codegen backend that compiles #[kernel]-annotated Rust functions to PTX through a Rust → Stable MIR → Pliron IR → LLVM IR → PTX pipeline, with single-source host+device compilation from one cargo oxide build command.
The post NVIDIA AI Just Released cuda-oxide: An Experimental Rust-to-CUDA Compiler Backend that Compiles SIMT GPU Kernels Directly to PTX appeared first on MarkTechPost.
NVIDIA researchers have introduced Star Elastic, a post-training method that embeds multiple nested reasoning models — at 30B, 23B, and 12B parameter scales — inside a single checkpoint, eliminating the need for separate training runs or stored model weights per variant. Built on the Nemotron Elastic framework and applied to Nemotron Nano v3, the method trains all three variants in a single 160B-token run, achieving a 360× token reduction compared to pretraining each model from scratch. Beyond training efficiency, Star Elastic introduces elastic budget control — a novel inference scheme that uses a smaller submodel for the thinking phase and the full model for the final answer — delivering up to 16% higher accuracy and 1.9× lower latency compared to standard budget control, while nested FP8 and NVFP4 checkpoints bring the full model family within reach of RTX-class GPUs.
The post NVIDIA AI Releases Star Elastic: One Checkpoint that Contains 30B, 23B, and 12B Reasoning Models with Zero-
Hyperscale cloud providers are doing what any aggressive buyer with deep pockets would do: purchasing enormous volumes of DRAM and high-bandwidth memory to feed AI factories, new cloud regions, and expanding platform services. By securing supply ahead of competitors, they lock in favorable terms and ensure their growth is not constrained by component scarcity. From their perspective, this is smart business. From the enterprise market’s perspective, it is something else entirely.
When the largest infrastructure providers absorb a disproportionate share of a finite supply of memory, prices rise for everyone downstream. Enterprises attempting to refresh on-premises servers, expand private clouds, or maintain hybrid architectures suddenly face a distorted market. Hardware lead times grow. Budget assumptions fail. Planned refreshes become much more expensive than expected. In some cases, the cloud begins to look attractive not because it is strategically superior, but because the economics
AI will help build the energy it needs. That’s the case U.S. Energy Secretary Chris Wright and NVIDIA Vice President of Hyperscale and High-Performance Computing Ian Buck made Thursday morning at the SCSP AI+ Expo. The 30-minute fireside chat, moderated by SCSP president Ylli Bajraktari, was called “Powering the Next American Century.” Their argument: American […]
SANTA CLARA, Calif. and CORNING, N.Y., May 7, 2026 — NVIDIA and Corning Incorporated have announced a multiyear commercial and technology partnership to dramatically expand U.S.-based manufacturing of the advanced optical […]
The post NVIDIA and Corning Announce Long-Term Partnership to Strengthen US Manufacturing for AI Infrastructure appeared first on AIwire.
MRC (Multipath Reliable Connection) is a new open networking protocol developed by OpenAI in partnership with AMD, Broadcom, Intel, Microsoft, and NVIDIA that improves GPU networking performance and resilience in large-scale AI training clusters by spreading packets across hundreds of paths simultaneously, recovering from network failures in microseconds, and enabling supercomputers with over 100,000 GPUs to be built using only two tiers of Ethernet switches.
The post OpenAI Introduces MRC (Multipath Reliable Connection): A New Open Networking Protocol for Large-Scale AI Supercomputer Training Clusters appeared first on MarkTechPost.
