To give it the fairest possible test I will take your concerns to an optimistic extreme.
Given that I can set my thermostat back roughly part time, and most people work 40 hrs, and someone who can blow $300 on a thermostat probably works at least 60 hours, I will triple my theoretical savings to $36/year. Given a zero interest loan that would pay for itself in a "mere" nine years. Google products don't last nine years before being shut down but I will error on the side of optimism and assume this is an outlier. The problem is the 29.99% credit card to buy the $300 thermostat means it really costs $90/yr to buy one... and now we're saving $36/yr instead of $12, but its still an order of magnitude or so too little to run a profit before it requires replacement.
Still not seeing it.
The target audience is people who can't do math. Or like greenwashing. Or like showing off to other people that they can throw away $300 on a thermostat, or at least obtain a loan to do so. There's nothing particularly wrong with that as a market, either. I read an online article from three months ago that they were selling 40K/month and assuming they get a revenue of $100/unit (rest goes to retailer, warehouse, shipping, etc) thats a revenue of about $50M/yr which isn't bad. A 10% profit on revenue on electronics would be good, so lets say $5M profit per year. Well, I wouldn't pay about $4B today for $5M next year, but aside from that little issue...
I don't claim the product, in itself, is foolish at all. As a "hey look at how much money I have, err, had" $300 product, its probably a better long term investment than a $300 cellphone or a $300 computer video card or even a $300 giant TV. Some of the mathematical / economic rationalization is hilarious but that's orthogonal to the foolishness argument.