daily · 9 nov 2026

NVIDIA opens an inquiry into Huang

The Token Times · Daily · Monday, November 9, 2026

The last 24 hours in AI, ML and tech.

1. NVIDIA board opens independent Huang investigation · A formal national security review reaches the leadership of a central AI chip supplier.

2. Dynamo deployment recipes for Korean open models · Four production inference recipes target B200 GPUs, with a vendor claim of up to 2x performance.

3. Rebellions and SK Telecom plan NVLink Fusion inference · GPU prefill and NPU decode could change inference economics; deployment is expected in 2028.

4. NVIDIA announces Point2 investment · AI interconnect funding targets the movement of data between chips and racks.

5. LG AI Research opens K-EXAONE-2.5 API access · New API access adds a Korean model developed with NVIDIA Nemotron datasets.

6. LNbits orders server shutdowns after attacks · An active attack wave prompts a project-wide stop order while security patches are pending.

7. Airlock launches versioned policies for coding agents · Repo-scoped tokens and file-level autonomy controls address enterprise agent deployment.

8. Six GPUs behind a consumer PC’s PCIe switch · A detailed local inference build shows ROCm and CUDA sharing a consumer host.