425: Corvex Launches Secure AI Model Weights, Bolstering IP Protection

Corvex, a company set to merge with Movano, announced the early availability of Secure Model Weights, a patent-pending solution for hardware-enforced AI inference protection.

Summary

• Corvex launched Secure Model Weights on March 12, 2026, a patent-pending solution for protecting AI model weights during inference on third-party GPU infrastructure.
• The solution addresses the 'cleartext gap' where model weights are exposed at runtime in traditional cloud security models, which can lead to intellectual property theft.
• It leverages Hardware-based Trusted Execution Environments (TEEs), including NVIDIA's Confidential Computing instructions and Intel Trust Domain Extensions (TDX), to ensure weights are decrypted only within the GPU's secure silicon boundary.
• The technology combines TEEs, Remote Attestation (cryptographically verifying host configuration), and Post-Quantum Key Exchange (ML-KEM/Kyber-768) to protect against current and future cryptographic threats.
• Corvex Secure Model Weights is built on an open-source foundation, utilizing the Confidential Containers (CoCo) project, allowing for vendor-neutral, community-audited security.
• The product targets frontier AI model builders and regulated enterprises (e.g., healthcare, finance, defense) that require verifiable, hardware-enforced protection for sensitive datasets.
• Movano Inc. (Nasdaq: MOVE) previously announced a definitive agreement on November 10, 2025, to combine with Corvex in an all-stock merger.

Sentiment

Score: 8

Explanation: StockSavvy.ai views this as a highly positive development, introducing a critical security solution for AI infrastructure that addresses a significant market need and enhances Corvex's strategic value, especially in light of its pending merger with Movano.

Positives

• Provides hardware-enforced protection for valuable AI model weights, safeguarding intellectual property during inference on third-party infrastructure.
• Closes the critical 'cleartext gap' vulnerability in traditional cloud security models, where data at runtime is typically exposed.
• Utilizes advanced technologies such as NVIDIA Hopper and Blackwell GPUs in Confidential Computing mode and Intel TDX for robust CPU and GPU-level isolation.
• Incorporates Remote Attestation to cryptographically verify host hardware and software configurations, preventing key release to compromised systems.
• Employs Post-Quantum Key Exchange (ML-KEM/Kyber-768) to protect against future cryptographic threats, ensuring long-term security for model weights.
• Built on an open-source foundation (Confidential Containers project), enabling independent verification and auditability of its security claims.
• Enables regulated enterprises and frontier AI model builders to securely deploy sensitive workloads on external infrastructure, expanding their operational capabilities without compromising security.

Risks

• Forward-looking statements contained in the press release are subject to a number of risks, uncertainties, and assumptions that may cause actual results to differ materially from those projected.
• There is no guarantee that future results of operations, financial position, projected growth, product capabilities, or customer deployment will be achieved as anticipated.
• Risks and uncertainties related to the proposed all-stock merger between Movano and Corvex are detailed in the registration statement on Form S-4 and the accompanying proxy statement/prospectus filed with the SEC.

Future Outlook

Corvex anticipates future growth in product capabilities and customer deployment, aiming to enable secure AI inference at production scale for frontier AI model builders and regulated enterprises. The company believes its solution will allow infrastructure partners to be chosen based on verifiable security, not just price and availability, thereby transforming the industry's approach to AI deployment.

Management Comments

• Seth Demsey, co-founder and co-CEO of Corvex: "Deploying AI should never require a trade-off between compute power and IP sovereignty. The industry has long tolerated a 'cleartext gap' where weights are exposed during inference, leaving the host with a structural window into your trade secrets. We are closing that window. By enforcing end-to-end encryption that terminates only inside the GPU's trusted execution environment, we ensure that the host is never in possession of the keys to the kingdom. Protection is no longer a policy choice; its a certainty."
• Jay Crystal, Co-CEO and Co-Founder of Corvex: "Model builders and security-conscious enterprises are now able to choose infrastructure partners based on verifiable security, not just price and availability. An open source foundation and owner-controlled key custody are what make that security auditable and trustworthy. We built Secure Model Weights so that the model builder never has to take our word for it. The math and the hardware speak for themselves."

Industry Context

StockSavvy.ai notes that the launch of Corvex Secure Model Weights addresses a critical and escalating concern within the AI industry: the security of proprietary AI models and sensitive data during inference on third-party cloud infrastructure. As AI models grow in complexity and value, and as regulatory demands for data privacy intensify, solutions offering hardware-enforced intellectual property protection are becoming indispensable. This strategic move positions Corvex to capitalize on the significant demand from frontier AI labs and regulated enterprises that have historically faced substantial barriers to adopting external compute resources due to stringent security requirements.

Comparison to Industry Standards

• Corvex's solution directly contrasts with traditional cloud security models that leave data exposed at runtime, addressing a 'cleartext gap' that many existing industry offerings may still exhibit.
• Unlike closed-source commercial alternatives, Corvex's open-source foundation (Confidential Containers project under the Cloud Native Computing Foundation) provides a higher level of auditable and verifiable security, potentially differentiating it from proprietary solutions offered by major cloud providers or specialized security firms.
• The integration of NVIDIA's Confidential Computing instructions and Intel Trust Domain Extensions (TDX) positions Corvex at the forefront of leveraging hardware-based Trusted Execution Environments, a standard that is becoming increasingly vital for high-security AI workloads.

Stakeholder Impact

• Shareholders (Movano): The successful launch of Corvex's new product enhances the strategic value and market positioning of the combined entity post-merger, potentially leading to increased shareholder value.
• Customers (Corvex): Frontier AI model builders and regulated enterprises gain access to a highly secure and verifiable solution for deploying sensitive AI workloads, enabling greater flexibility and trust in third-party infrastructure.
• Employees (Corvex): The product launch reinforces the company's engineering leadership and market relevance, potentially boosting morale and attracting talent.

Next Steps

• Investors and stockholders are urged to read the Form S-4, the Proxy Statement, and any other relevant documents filed by Movano with the SEC regarding the proposed merger with Corvex.
• Corvex Secure Model Weights enters early availability on March 12, 2026, with interested parties able to learn more or request early access via www.corvex.ai/confidential-computing.

Key Dates

Recommendation

The launch of Corvex Secure Model Weights represents a significant technological advancement addressing a critical and growing need for security in AI inference. This product enhances Corvex's strategic value and competitive positioning within the high-growth AI infrastructure market. Given the pending all-stock merger with Movano, this development is highly favorable for Movano shareholders, as it strengthens the combined entity's future prospects and intellectual property portfolio, making the stock a strong buy for investors seeking exposure to secure AI infrastructure.

Keywords

AI security, model weights, confidential computing, GPU infrastructure, trusted execution environments, NVIDIA, Intel TDX, post-quantum cryptography, open source, enterprise AI, Movano merger, data protection