TechPrufer
Sign inGet started

Blogs›NVIDIA›Hardware-Rooted AI Security with Confidential Computing

Hardware-Rooted AI Security with Confidential Computing

Hardware-Rooted AI Security with Confidential Computing

1
posts
2026

This feature thread tracks the evolution of hardware-rooted AI security, focusing on NVIDIA Confidential Computing (CC) to protect data, code, and model integrity during inference with minimal performance overhead. Initial efforts focused on the challenges of securing AI workloads, leading to the development of CC integrated into Blackwell GPUs. Subsequent developments have demonstrated near-native inference performance (up to 98%) through optimizations in secure work submission, bandwidth management, and integration with inference frameworks like FlashInfer and SGLang. This includes advancements in remote attestation via NRAS, NVLink encryption, and hardware-level root of trust, enabling secure, production-scale AI deployments.

2026

Hardware-Rooted AI Security That Won’t Slow You Down | NVIDIA Technical Blog

7/2/2026

This post introduces NVIDIA Confidential Computing (CC) as a hardware-rooted AI security solution. It details the integration of CC into Blackwell GPUs, including fused private signing keys and NVLink encryption. The post explains the remote attestation process using NRAS to verify workload integrity before execution. Benchmarks on HGX B300 with the Qwen 3.5-397B-A17B-FP8 model demonstrate minimal performance overhead (under 8%) for throughput and per-token latency. It also highlights optimizations in inference frameworks like FlashInfer and SGLang (CC-safe autotuning, async D2H copy worker, piecewise CUDA graph support) that mitigate performance impacts.