I'd say neuro tech for gaming/research/medical imagining is easy and high. Most of the market in the consumer space is trying to sell "calm" and shit (while people are spending a couple of hundreds on consoles + games vs vc funded hype machines repackaging headphones), researchers are building their own systems and using it for papers that at a cost of <$5k vs the $100k system I used last in a lab, but whose projects have caveats like "we didnt care about price so we paid $600-$800 for an fpga + peripherals when we could have spent $80, we paid $900 for 3 axis accelerometer and microprocessor when could have spent $50 but meh, our grant monies pay for this!", and the state of the art techniques for getting under 1mm in spacial and sub millisecond precision have been used in radar systems for about 80 years (and even lower spacial precision when you combine recent techniques like fast ica, with methods like jacobi-davidson, and shit tonnes of awesome libs like armadillo, superlu, openblas to parallelize the problem for real time implementation over many (~200) cheap microprocessors that can fit within 70mm x 10 mm x 70mm)… thats not to say theres not work to be done, but its deff not limited by the technology and costs…
I'm staying the hell away from the reality distortion field that is the %80-%90 of the industry right now lol