Our brain has some 100 billion neurones and uses roughly 20W of power. That is really really low power (although it represents 20% of the total power we use).
By comparison, the Apple A8 chip, one with the lowest power consumption on the market, uses roughly 2W to power its 2 billion transistors. If we were to associate one transistor to one neurone the cluster of A8 will consume 5 times as much as our brain. However, we cannot equate one transistor to one neurone since we need to take into account the connectivity and pre-processing aspects, and here a brain has between 100 trillion to 1,000 trillion synapses whilst a transistor has just ... 1. So if we were to compare the A8 in terms of synapses we should have a power consumption in the order of hundreds of KW versus the 20W of our brain.
Clearly the comparison is difficult but it is obvious that a brain uses much much less power than a chip.
Scientists are at work to create chips that can mimic the processing of neurones and neurones circuits (like Synapse, by IBM who aims at, eventually, mimic 1 trillion synapses uses "just" 4kW of power, look at the infographic) but the "power" required remains a major stumbling block.
This is why this news coming from Pohang University of Science and Technology in South Korea (POSTECH) is so important.
A team of researchers have shown the possibility to create an artificial synapse using organic nano-fibers (ONF) with a power consumption that is one tenth of a natural synapse (1 femto joule vs 10 femto joules needed by a natural synapse to perform a single synaptic event). The ONF can be precisely organised to mimic neuronal connection of a brain and can be produced in huge quantity. This, according to the researchers, can make possible the development of a neuromorphic electronic chip that can be used for storing -and processing- huge volumes of data.
This evolution,however, is still a dream although it may come true in the next decade.