There's plenty of non ML software engineering to be done. Anecdotally a large proportion of interns want an "ML project", but only a small percentage of teams looking for interns are offering one.

Too many people going into ML could skew the supply/demand into making it a worse job option (more work, less pay), like game programming or academia.

I got an offer to do ML on video data recently, but turned it down to do some non-ML software eng. instead, for double the money. The ML work would be way cooler, but it would at a startup (=stressful) and double the money means half the time to reach FIRE.

The caveat is that I've worked on ML in the past and I think that the work is maybe less intellectual than software engineering - with complex enough models they become impossible to understand and you start to just try out ideas based on random intuitions. The thing I mostly like about it is the ability to use math and more independent style of work - no scrum, less need for cooperation with other team members etc..

I started in ML about 7 years ago so well before the hype and back then very few people wanted to be ML engineers.

What's happening at least in Australia now is that contract rates (a good indicator of the supply/demand ratio) has halved for ML engineers. Which means (a) a lot of people want to be ML engineers and (b) there aren't that many jobs for them.

ML engineer is a super boring job content-wise and has insane outside pressure. It's about building data pipelines, the ugly grunt work. ML/Data Scientist is the interesting job. Usually Data Scientists view ML Engineers as replaceable drones that don't understand anything interesting and do the boring part of the job for 2-3x less than they do. The only advantage of ML Engineers is that AutoML is unlikely going to replace some dirty work but might endanger outdated Data Scientists.

Well, you could save money and use it to feed yourself while learn new skills when you're out of a job. That works too.