Hardware-based trusted execution environments (TEE) like Intel’s Software Guard Extensions (SGX), AMD’s Secure Encrypted Virtualization (SEV), and Nvidia’s Confidential Computing technologies were designed to protect sensitive data and code even if the operating system is compromised. These on-chip enclaves offer strong software-level isolation, making them a cornerstone of modern secure computing. However, recent developments show that their security is increasingly under threat from low-cost, physical attacks that bypass their core assumptions. While TEEs are resilient against software-based exploits—such as root access, malware, or hypervisor attacks—they are not immune to physical access. Attackers with direct access to a device can now deploy simple, inexpensive techniques to extract data from enclaves. These include side-channel attacks, power analysis, and fault injection, all of which can be carried out with off-the-shelf equipment costing just a few hundred dollars. For example, researchers have demonstrated that by monitoring power consumption or electromagnetic emissions during enclave operations, it’s possible to infer secret keys and data. Similarly, fault injection attacks—using laser pulses or voltage manipulation—can disrupt the normal operation of a chip, causing it to leak information or behave unpredictably, potentially exposing encrypted data. The threat is further amplified by the fact that many of these physical attacks do not require advanced lab equipment. In some cases, attackers can use basic tools like a microcontroller, a power supply, and a simple oscilloscope to perform successful exploits. This accessibility means that even non-state actors or determined individuals can compromise what was once considered a gold standard in hardware security. As a result, the security model of TEEs—relying on the assumption that the physical environment is trusted—is being rapidly eroded. The growing number of real-world incidents, including the exposure of encryption keys and private data in cloud environments, has raised serious concerns about the long-term viability of these technologies. While chipmakers are responding with new countermeasures—such as enhanced shielding, dynamic power management, and hardware-based anomaly detection—these solutions are often costly and not yet widely deployed. In the meantime, the security of on-chip enclaves is no longer a given, especially in environments where physical access is possible, such as data centers, shared cloud infrastructure, or devices that leave the controlled supply chain. The takeaway is clear: while TEEs remain valuable for protecting against software threats, their effectiveness is severely diminished when physical access is involved. As physical attacks become more common and easier to execute, organizations must rethink their security strategies and consider additional layers of protection beyond hardware enclaves.