And the lesson that I hope the maths teacher draws from this is that doubling each time (ie 2x) grows faster than powers (ie xa), for any finite a, given a large enough x.
Imagine a slightly different chessboard, where on the first square you put one grain, on the second 4 (2x2), on the third 9 (3x3), on the fourth 16 (4x4) and so on: the nth square gets n^2 grains.
The 64th square gets 64x64, or 4096, which is rather less than 1.86 x 10^19 (that's 18 followed by eighteen zeros).
The next board could cube the number: the first square gets 1, the second gets 8 (2x2x2), the third gets 27 (3x3x3). Then the 64th square gets 64x64x64 = ~250 thousand. Still rather less.
The next board could multiply the number by itself four times: 1, 2x2x2x2=16, 3x3x3x3=81, 4x4x4x4=256 and so on. The 64th square gets ~17 million.
How long would you have to keep doing this before the 64th square gets more than 2^64?
The answer is "the eleventh board" as 6411 (64x64x64x64x64x64x64x64x64x64x64x64) is indeed greater than 264 (2x2x2...total of 64 2s). (exercise: show this is true, without working out what 264 or 6411 are).
But all you need then is a few more squares: 267 is 1.4x1020, while 6711 is is "only" 1.2x1020, and off it goes, with the doubling board running ahead.
OK, you say, let's do another board, this time going 1, 4096, 531441, 16777216 (ie 112, 212, 312, 412...). Now we'll beat the doubling board, won't we? Well, until the 75th square, at which point doubling pulls ahead again.
It's another interesting exercise (good GCSE student, I assume/hope) to work out how many squares you have to go before the doubling board outstrips the "to the power of" board, without using a calculator (or log tables), for any given power. By square 128 you would need to be 219, 319 and so on to be ahead, but you'd be behind within a few squares. By square 256 232, 332, 432 could stay with the doubling (because 2256 = 256^32), but on square 257 the doubling gets ahead again.