You're right that sensored motors add a bit of cost. Then again, motors with encoders are a Good Thing (tm) so perhaps something smart can be done around that?
For affordable sensored brushless, the Chinese hobby people are hard to beat:
Sensorless of course go down to half the price or lower. With a high pole count, you can establish the stall point yourself, and then dead reckon the motor, basically treating it as a multi-phase stepper. 24/22 motors are reasonable affordable, and have more precision than a typical bipolar stepper.
When it comes to BLDC driver chips, I think Texas Instruments pretty much owns
DRV8312 (Edit DRV8332 even better) seems largely self-contained, and would pair OK with smaller motors.
DRV8305 looks good (as a gate driver) for example. In general, my theory is that the small cost differences for power MOSFETs don't favor the "10A and below" crowd; I like chips like Vishay SISA72DN (40V, 60A) or perhaps BUK9Y14-40B. At least 40V rating is nice to avoid failure when seeing inductive backlash, and 30A and up rating is needed because of the high amps needed for torque if you don't have a gearbox. Plus, lower Rdson means less board area / heat sinking needed even for lower currents, so saving 50 cents (or even 3 dollars) on total BOM cost and losing out on this seems counter-productive IMO. Super low Rdson is great not because I'm driving kilowatts of motors, but because the board/cooling becomes simpler for the motors I do drive I imagine the real cost is in you guys' time and warehouse rental and shipping and such.
The main problem I have with brushless is that I need huge amperages to match the torque of brushed motors, which is a drain on battery. A gearbox (planetary, or even micro cycloidal) would be nice. Or, alternatively, if there is sensored control, chopping current such that it meets the desired position/velocity and no more. (I think some of the driver chips have a separate enable bit that could perhaps achieve this.) And ultrasonic switching frequency. Because quiet matters and I want it all
I don't know how big your volume is, and how much clout you have with various manufacturers. Given how simple a basic brushless motor is, if you could build a motor-with-absolute-encoder and use that encoder as an input to a controller, that might make for a very nice setup. (4096-16384 step magnetic encoders like the AS5048 seem like a reasonable option.)
There are lots of little pieces to this equation, and the delineation is not as clear as it is with brushed motors / controllers. Although steppers seem to be doing fine, and BLDC isn't really that much more complex. There really ought to be some sweet spot here to exploit.