Let’s face it: cloud is someone else computer with fancy APIs. Whether one of its canned features can solve a problem or not is best assessed by a team with full context over a problem. Cloud solution architects in practice are distant from such, ending up “enabling” features based on buzz and/or preferences.
A general enterprise-wide architecture direction leads to Big Design Up Front (BDUF) issues. Its fine to have a preferences of which technologies to consider first; yet that should be an open-ended, bottom-up effort. In that sense, Staff Engineering is a better approach.
Another aspect are cloud platform certifications. They focus on a learn-by-heart strategy where you need to remember all service tiers and options. In practice you never remember those, and even if you do it may not be applicable anymore due to how fast those offerings change.
In general cloud architects are a bad investment for companies. The sense of tight control over what’s available for internal developers is offset by an echo chamber between those that consume those services and those that make decisions about the offerings.