Ok, this isn't rigorous at all (obviously), but it seems to me that if the size of the gap continuously grew, but fluctuated randomly, you would still have an infinite number of primes close together even though the average distance between them never stopped increasing. They would become fewer and fewer, but never stop, and hence would be infinite.
Not doubting the guy's work, but I'm doubting the summary's "the gaps between consecutive numbers don't keep growing forever."