The conventional answer was that microelectronic circuits are far less integrated than biological circuits, but that is less and less true with microelectronic circuits in the 10 nanometers when a dendrite is 1 micrometer.  But microelectronic circuits are 2D and they pay a huge price in interconnections, and the neurons feed locally, no need for power wires. The quick answer would be to have 3D chips (it was also the quick answer four decades ago), but it would involve a new processing technology and as microelectronic plant costs several billions Dollars, nobody will try that soon if there is no strong incentive.

I always find these neural network projects so fascinating, I wonder if there ever will be a time when a "real" synthetic type cellular network can be made to respond and learn in a similar manner like our brains do...but at far more geometrical proportions?