Is a self-signed certificate inherently less secure than a CA-issued one?

Is a self-signed certificate inherently less secure than a CA-issued one?

The common advice — "never use self-signed, it's insecure" — conflates two distinct properties. A certificate does two jobs: it carries a public key for the key exchange, and it carries an identity assertion a relying party can verify. The cryptographic strength of a TLS (Transport Layer Security) session depends on the key and negotiated cipher suite, not on who signed the certificate. A self-signed RSA-3072 or P-256 certificate produces exactly the same handshake confidentiality as one from a public CA (Certificate Authority).

What self-signed certificates lack is third-party identity binding. The signature only attests "this key signed itself," so a relying party with no prior trust anchor cannot distinguish it from an attacker's. That is an authentication gap, not an encryption gap.

The distinction matters operationally. For internal service-to-service traffic where you control both endpoints and pin the certificate (or run a private CA), self-signed or private-CA certificates are entirely appropriate — see RFC 5280 §6 on path validation, which is what public trust automates.

— Encryption: identical
— Identity: absent unless pinned or pre-distributed

Further reading: RFC 5280, §6 (Certification Path Validation).
Bottom line: "insecure" is the wrong word. Self-signed certificates lack delegated identity verification, which only matters when the client has no out-of-band way to trust the key.