I find that names and terms used in relation to Trusted (Trustworthy, Confidential,..) Computing are highly interchanged and thus creating confusion for laymen as am I.
Trusted Computing has been around since the 90's and the idea is being tried to be kept up to date with today's standards but the initial idea is that a computer system can be trusted (work in a secure way) when a hardware (predominantly) root of trust is present thanks to which the system can attest to its current state, use it's secrets to encrypt data etc.
A system with just a TPM chip as a security measure does not add much protection for data in use, but rather for data in storage.
Today's security standards call for different measures then for those defined 30ish years ago (though updated) and thus TEEs are being widely deployed across systems (Apple's SE, Intel's SGX, ARM's TrustZone, many different frameworks for cloud (Enarx,..) etc.) and many of them support most, if not all, or TPMs function (like attestation, DAA, ...).
Though, my assumption is for every TEE a hardware root of trust must be present either in the form of a TPM or a simple fuse or eMMC storage containing a secret key from the manufacturer.
Am I right in assuming that the role of a TPM chip (excluding TPM-like features of TEEs) has becoming obsolete as new generation technologies (CPUs) already rely on many Trusted Computing principles? Is it sufficient to rely on a different ways a hardware root of trust can be implemented in different TEEs and call it a Trusted Computing when TEE implementations follow TCG's standards? Is the way TEEs work even compliant with Trusted Computing? Am I mixing too much together?
