Approximately one year ago technology manufacturers acknowledged the end of Moore's law (the trend of processing power doubling every two years). Indeed for several years before that announcement it was necessary for all the manufacturing giants to collaborate together in order to keep up with Moore's law. And current microprocessor technology has already hit a physical limitation due to heat production. You may have noticed that suddenly computers & laptops all started having dual-core processors, this was the work around manufacturers came up with to keep up with Moore's law despite the physical impossibility of making more compact chips.
With this in mind, consider the power of the current top of the line chips. The largest commercially available single-chip processor packs in a whopping 7.2 billion transistors. But how does that compare to the human brain?
The human brain contain approximately 100 billion neurons (10^11), more than ten-times the number of transistors in these approaching-the-laws-of-physics chips. But a single neuron is way more complex than a single transistor. A transistor can only have two states: on and off and have very limited inputs & outputs. Whereas neurons can send & receive hundreds of different chemical signals with thousands of other neurons. So let's consider instead that a single transistor can service as a single synapse (interface where two neurons communicate with each other). Well it's been estimated that there are approximately 1 quadrillion synapses (10^15). So we're talking >100,000 top of the line processors to model a single human brain.
But again a single synapse is much more complex than a single transistor as it consists of hundreds of different proteins and thousands of copies of each one not to mention all the proteins involved in transmitting signals between different synapses. As a rough approximation the human brain weighs ~1,500g of which 75% is water. Leaving ~375g of not-water which we'll assume is at least 50% protein (188g). A single protein molecule weighs approximately 10^-19 g, So there are approximately 10^21 protein molecules in a typical human brain.
Whether it is necessary to model each of those molecules in order to recapitulate human consciousness is currently not known. Many would argue that it is only necessary to model the network of neurons to recreate human consciousness, but then others argue the quantum state of particular proteins hold the key to consciousness. But regardless of whether it is 10^15 synapses or 10^21 proteins or a combination of both, it will take an entire server farm to imprecisely model a single human brain for the foreseeable future.
Using estimates of Google's computer power by Randall Monroe and the extremely over simplified 1 transistor per synapse model, then all of Google could model around 2,000 human brains. But that's not counting back-ups because in the same article it was estimated Google has a drive fail every few minutes. So each of our modeled human brains will have a failure every two days, and consume hundreds of kilowatts of electricity. Suddenly, a blob of goo the size of a cabbage which lasts for 60+ years using only 2,000 kJ of energy per day (0.02 watts) is looking pretty good.....
