The fact that tampering resistant (i.e. make tampering more expensive / complex) processors like developer.arm.com/products/proce require signing an NDA to learn details about the mitigations just goes to show how little faith the designers of the hardware have in their own creations.
Conversation
Resistance to physical attacks can have value, but it's a losing battle. If it's a decent design, they'd publish details on how it works to promote it. A good lock doesn't rely on an attacker not knowing how the locking mechanism works. It relies on per-device hardware secrets.
Replying to
Anti-tampering for computing hardware is primitive and relies heavily on design secrecy and piling on complexity to increase costs. It's just like piling on weak software mitigations not providing fundamental security properties. It can raise costs but it's hard to quantify that.
1
6