In the framework of the privacy-preserving concerns of the SOTERIA project, and concerning future Machine Learning (ML) implementations, we explore hardware-based privacy computations for secure and private data storage and ML computations. Current supported hardware on cloud and consumer devices most notably include Trusted Execution Environments (TEEs), which have emerged as a critical technology in modern computing, offering hardware-enforced security measures to protect sensitive data and computations from potential threats. This report presents a comprehensive review of TEEs, focusing on the advantages, vulnerabilities, and implementations of prominent TEE technologies, and the state-of-the-art on TEE-based methods for privacy-preserving sensitive data storage and ML operations. Finally, we provide a discussion on these commercial implementations, methods, possible implementations for improving the security and privacy on the SOTERIA platform for current and future applications, and our conclusion.