Bandwidth is a major hurdle in the development of Brain-Machine Interfaces (BMIs). Currently, the maximum number of neurons that can be recorded at once is a couple hundred, but a million is needed to create a wide range of useful BMIs.
Moore’s Law suggests that this could be achieved by the end of this century, but implantation is another major hurdle. Invasive brain surgery is expensive and limited, so a non-invasive solution is needed. In addition, the device must be biocompatible, wireless, and powerful enough to send and receive data.
Finally, a breakthrough in machine learning is needed to decode the code of millions of simultaneously chattering neurons.
