Additionally, could you elaborate on the specifics of situation where you used, or were exploring the possibility of using, these technologies?
I was first exposed to AWS Lambda when working on a scraper that needed to operate in a few different regions. Its scale was relatively small and it would only need to do its thing when a scraping job was manually submitted by one of the users.
Initially, the Lambdas only interested me because they could easily provide my "scrapers" with a nice and reliable way to access rotating IPs in the several regions I need my script to work in. Shortly, I realized that it could offer me quite a bit more than that: it could abstract away the troubles of creating the architecture needed for running distributed tasks in the background.
Scalability was basically not a concern at all in my case, so Lambdas weren't really adding much value in that regard. However, with the isolation of functions into independent units that could asynchronously be invoked from my code abstracted a lot of infrastructure work. I no longer needed to code a background worker, I didn't need to care about queueing up jobs, and if I ever needed to scale larger - I was already there.
Hooooowever, that's not a complete picture. Yeah, I was kind of able to set up everything in a way that would do all the things I mentioned but it was only after a bunch of cursing and frustration.
The documentation sucked. Packaging dependencies for deployment sucked. The deployment itself sucked. Figuring out how to invoke it in a way that I wanted sucked. In the end, although the service was one of a kind, the unintuitive experience of working with it left me wanting for an alternative that does the same thing... but well.
My experience is definitely biased, I was not looking for something that would solve my scalability issues, I was looking for a solution that could abstract the architectural headaches. In that regard, it failed me.