... And this is why a lot of attendings tell their students/residents to specialize.
Keep in mind, academicians usually are clueless about what happens in the real world outside their bubble. I'd take their advice with a grain of salt.
You can still find many, many jobs in neurology without a fellowship.
Also, fearing midlevels should never be the reason to subspecialize. They are, literally, in every field. Plenty of them are in surgical fields. Yes, they don't do surgeries yet, but they do a lot of the other work that surgeons should be doing (following up on postop pts, writing notes, etc). This increases surgeon's efficiency and will eventually lead to decrease need.
Very few subspecialties in Neurology, and in medicine in general, that make good financial sense. When you decide on pursuing a fellowship, you are sacrificing attending salary for X amount of years. Therefore, in neurology, unless you're going to do NIR or interventional pain fellowship, no fellowship will result in a meaningful boost to your income to offset the opportunity loss of doing a fellowship.
Look at the following two extremes to get the idea:
John goes into FM, finishes 3 years training, and gets a job paying 230k right out of residency
Jane goes pursues a training in interventional cardiology (8 years long). Gets a job paying 500k right out of fellowship
During the 5 years of the extra training that Jane endured, Johan produced $1,150,000 while Jane produced $300,000 (60k/year in fellowship).
This puts Jane $850k behind John. Given the gap between their salaries, it'd take Jane 3-4 years after her training to catch up to John.
Keep in mind, I didn't account for the difference in work hours during fellowship and beyond. From what I observe at my institution, cardiology fellowship regularly work 80+hrs/week. Most FM docs work less than 45hrs/week. Therefore, if we want to adjust for this, the opportunity loss Jane would endure can be much more than what the calculations above showed.