Duality Technologies Enables Secure GenAI Workflows on NVIDIA GPUs

ByRicardo

Duality Technologies, a frontrunner in privacy-enhancing applied sciences and safe knowledge collaboration, in the present day introduced assist for Google Cloud’s Confidential Computing portfolio, together with NVIDIA GPU-powered confidential digital machines on Google Cloud, enabling large-scale secured AI workloads reminiscent of LLM coaching and inference.

With this launch, the Duality Platform now helps GPU-backed LLM inference and encrypted Retrieval-Augmented Generation (RAG) inside trusted execution environments (TEEs) – a major efficiency leap from earlier CPU-only assist.

Customers can run a safe generative AI workflow on NVIDIA GPUs with Google Cloud Confidential Computing, now that includes end-to-end safety towards knowledge leakage powered by Duality. They can even mix full-stack knowledge confidentiality with NVIDIA H100 GPU efficiency, unlocking confidential AI use circumstances that had been beforehand impractical because of latency and throughput constraints.

“This adjustments the sport,” stated Dr. Alon Kaufman, CEO and Co-Founder of Duality Technologies. “Our prospects can now run privacy-preserving AI with LLMs at manufacturing scale. With GPU acceleration, the efficiency bottlenecks of safe computing are gone-making safe LLM coaching and inference sensible.”

The new functionality is constructed on Google Cloud’s Confidential Space and Confidential NVIDIA H100-powered confidential VMs, with assist for Intel TDX and Cloud KMS integration. Duality has efficiently validated the platform working a Mistral-7B mannequin utilizing encrypted vector RAG (through Faiss) in a completely confidential pipeline.

“With Confidential GPUs, organizations can course of delicate AI workloads completely inside trusted execution environments with out giving up efficiency,” stated Nelly Porter, Director of Product Management, Google Cloud. “Pairing NVIDIA H100-powered confidential VMs with Duality’s encrypted workflows permits LLM coaching and inference to occur at scale, with end-to-end safety from knowledge leakage.”

Key Highlights:

  • GPU Support for Confidential AI: Run safe LLMs and encrypted RAG on Confidential NVIDIA H100s
  • Scalable Performance: Orders-of-magnitude sooner runtimes vs CPU-only workloads
  • Enterprise-Ready: Meets the wants of regulated industries, protection, healthcare, and AI-native corporations
  • Seamless Cloud Integration: Now obtainable through Dynamic Workload Scheduler in Google Cloud’s Confidential Space

Until now, confidential AI was restricted to CPU-only environments – appropriate for fundamental testing, however inadequate for the calls for of large-scale AI. With the arrival of Confidential GPU as a part of the confidential computing portfolio, Duality prospects can now run each LLM coaching and inference securely inside Trusted Execution Environments. This breakthrough allows high-throughput, privacy-preserving AI workloads that had been beforehand unimaginable to execute – unlocking new prospects throughout industries and use circumstances.

This functionality is initially obtainable on the Google Cloud Confidential A3 digital machine kind in preview, with broader rollout anticipated later this yr.

To study extra go to dualitytech.com

The publish Duality Technologies Enables Secure GenAI Workflows on NVIDIA GPUs first appeared on AI-Tech Park.

Similar Posts