Duality Technologies has announced support for Google Cloud's Confidential Computing portfolio, including NVIDIA H100 GPU-powered confidential virtual machines. This integration marks a significant advancement, enabling large-scale, secure AI workloads like LLM training and inference within fully protected trusted execution environments (TEEs). The move eliminates the performance bottlenecks that previously made confidential AI impractical for production-scale applications, offering a powerful solution for regulated industries.
Quick Intel
Duality Platform now supports Google Cloud's Confidential NVIDIA H100 GPUs.
Enables secure LLM training and inference within trusted execution environments.
Provides orders-of-magnitude performance gains over previous CPU-only confidential AI.
Supports encrypted Retrieval-Augmented Generation (RAG) for private data use.
Validated using a Mistral-7B model in a fully confidential pipeline.
Meets the stringent data protection needs of healthcare, defense, and regulated sectors.
Until now, confidential AI was largely confined to CPU-only environments, which were sufficient for testing but could not meet the demands of large-scale model training and inference. By integrating with Google Cloud's Confidential Space and Confidential NVIDIA H100-powered VMs, Duality has overcome this critical limitation. This allows customers to combine the raw performance of NVIDIA's H100 GPUs with full-stack data confidentiality, making high-throughput, privacy-preserving AI workloads a practical reality for the first time.
This technological leap unlocks a new frontier of use cases for sectors that handle highly sensitive data. Organizations in healthcare, defense, and financial services can now run secure generative AI workflows and encrypted RAG on protected data without the risk of leakage. This enables them to leverage the power of LLMs for innovation and insight while maintaining strict compliance with data governance and privacy regulations, a balance that was previously difficult to achieve.
Company leadership and its cloud partner recognize the transformative impact of this development. "This changes the game," said Dr. Alon Kaufman, CEO and Co-Founder of Duality Technologies. "Our customers can now run privacy-preserving AI with LLMs at production scale. With GPU acceleration, the performance bottlenecks of secure computing are gone—making secure LLM training and inference practical." Nelly Porter, Director of Product Management at Google Cloud, added, "Pairing NVIDIA H100-powered confidential VMs with Duality's encrypted workflows allows LLM training and inference to happen at scale, with end-to-end protection from data leakage."
The collaboration between Duality Technologies and Google Cloud represents a pivotal moment for secure, enterprise-grade AI. By bringing GPU-level performance to confidential computing environments, it removes the last major barrier to the widespread adoption of privacy-preserving AI. This empowers enterprises to confidently deploy large-scale AI solutions, knowing their sensitive data and models are protected throughout the entire lifecycle, from training to inference.
About Duality Technologies
Duality is a pioneer in privacy-enhancing technologies (PETs) and secure AI collaboration, enabling regulated industries and governments to leverage sensitive data across organizational and jurisdictional boundaries.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.