In 2010 the average computer was calculating at 100 billion calculations per second, which is more computational power than we had in the entire US Space programme in the 60's and 70's. In 2023, eight years from now, the average thousand dollar computer that you will go down to the store and purchase is calculating at 10 to the 16 cycles per second. That's just a number unless you speak to a neurophysiologist who tells you it's the rate at which your visual and auditory cortex does pattern recognition. So what happens when a thousand bucks buys you the computational power of the human brain? 25 years later a thousand dollar computer will buy you the computational power of the whole human race.
Speaking as a long-term transhumanist, it's unknown whether or not progress in computing (i.e. Moore's law) will follow the same trend after 5-10 years from now. We are rapidly approaching the limits of conventional silicon-based technologies which have formed the basis of our exponential trend since the 1950's. Kurzweil argues that once we reach this wall, another paradigm will simply take over in the same way ICs replaced transistors which replaced discrete transistors which in turn replaced vacuum tubes but the alternatives that we've got so far seem to be far away from commercialization.
That isn't to say that it will never be achievable. I'm sure it will happen eventually. The very existence of our brain is proof that the barriers won't be insurmountable.