Stories

  1. Image: NVIDIA Newsroom

    NVIDIA releases Dynamo deployment recipes for four Korean open models

    NVIDIA has released Dynamo-optimized deployment recipes for four Korean open models: SK Telecom's A.X K2, LG AI Research's K-EXAONE 2.0, Upstage's Solar Open 2 and Motif Technologies' Motif-3. Announced at AI Day Seoul, the recipes are tuned for B200 GPUs, and NVIDIA claims up to 2x performance without naming the baseline or workload. Dynamo is NVIDIA's open-source distributed inference-serving framework, and the recipes are meant to speed the models into services funded under Korea's AI for All initiative.

  2. Image: NVIDIA Newsroom

    NVIDIA, Rebellions and SK Telecom plan R200 NPU racks with NVLink Fusion for Korean sovereign AI

    NVIDIA, Rebellions and SK Telecom announced a collaboration on sovereign AI inference infrastructure in South Korea built around Rebellions' next-generation R200 NPU. The R200 will connect within racks via NVIDIA NVLink Fusion on the MGX rack architecture, with NVIDIA GPUs handling prefill and NPUs handling decode. SK Telecom will deploy the systems in its AI data center, with deployment expected in 2028. R200 specifications, per-NPU NVLink bandwidth and performance results have not been disclosed.

Topics

Publishers