Corvex Pioneers Confidential Computing Deployment for AI with NVIDIA HGX Systems

Corvex Leads in Confidential Computing for AI on NVIDIA Systems

On March 3, 2026, Corvex, Inc., an innovative player in AI cloud computing, made headlines by announcing the successful deployment of confidential computing on NVIDIA HGX™ B200 systems. This achievement positions Corvex as one of the foremost companies to offer verified and secure AI deployment solutions on advanced GPU infrastructure, catering to organizations that rely heavily on data sensitivity and AI model integrity.

Corvex specializes in GPU-accelerated infrastructure that is tailored to handle AI workloads. The new deployment leverages encrypted communication across the NVIDIA NVSwitch and NVIDIA NVLink fabrics, which is pivotal in ensuring data confidentiality and integrity while AI processes are in operation. With Intel Trust Domain Extensions and NVIDIA Confidential Computing, the combination creates a formidable environment where end-to-end AI security is achievable. This means that not only is sensitive data protected when at rest or during transmission, but importantly, also while it is actively being processed in memory.

As businesses increasingly shift toward mission-critical AI implementations, the demand for assurances regarding runtime security has surged. Corvex addresses this need by enabling confidential computing that significantly reduces the risks associated with data in use. This approach integrates hardware isolation and cryptographic attestation, which collectively provide verifiable proof of platform integrity during active workloads.

The significance of remote attestation cannot be overstated; it gifts organizations with cryptographic verification that their system configurations—including hardware, firmware, and drivers—remain uncompromised even in real-time operations. Such validation is essential, particularly in industries with stringent regulatory demands and where protecting intellectual property in AI workloads is non-negotiable.

Seth Demsey, co-CEO of Corvex, emphasized this point, stating, "In production AI, security is only trustworthy if it can be independently verified. Confidential computing makes trust at runtime measurable, employing innovative hardware isolation and cryptographic mechanisms across systems. This ensures that sensitive AI models and confidential data remain secure during processing."

The deployment of confidential computing on NVIDIA HGX™ B200 systems also marks a significant milestone in balancing the historical dichotomy between security and performance. The near-native performance provided by this infrastructure allows organizations to confidently execute sensitive applications without sacrificing speed or efficiency.

By facilitating the secure deployment of AI workloads that are enterprise-grade, highly regulated, or sovereign, Corvex reinforces its commitment to providing comprehensive security strategies that stand up against current threats in the AI landscape. This development also supports multi-tenant environments where security and compliance must be validated continuously during operation.

About Corvex

Founded as a leader in AI cloud computing solutions, Corvex is devoted to delivering GPU-accelerated infrastructure for complex AI workloads. By harnessing advanced compute clusters, high-throughput storage, and a layered architecture, Corvex ensures its clients benefit from enhanced security and consistent performance at scale. Previously, on November 10, 2025, Corvex announced a merger agreement with Movano Inc., further enhancing its capacity to innovate and serve the AI community.

This breakthrough in confidential computing not only advances Corvex's objectives but also sets a precedent for security protocols across the industry. As AI technologies progress, understanding how to protect and substantiate model integrity will become increasingly critical for organizations across varied sectors.