a few points to correct
one coin is not produced per block. meaning block 2,100,000 will not be the block with the final 'created' coin reward

there are 210,000 blocks per halving. and there are 32 halvings. meaning ~block number 6,930,000

after that, its just pure transaction fee as the reward
however thats in year 2140~.. and also for maybe 100 years prior to that(2040) people wont be concentrating on the block reward 'created' coins. as the amounts would be less than a single coin from 12 years time. where as the transaction fee's would be over 1btc(as they are now)

think of it as created coin as salary and fee as commission.. and your pay structure just shifting whereby the commission oversteps how much salary you get untill you just forget about caring about the salary because you are getting most of your income from commission. and then one day when you are 140years old.. your earnings is just pure commission

https://en.bitcoin.it/wiki/Controlled_supply#Projected_Bitcoins_Long_Term

...
also technically.. there is no btc in bitcoin. there is no decimal point
yep. thats right you will find no 8decimal number in any raw blockdata or raw tx
the 'btc' value is a division of satoshi amount100,000,000 that is done purely at the user interface level

at binary level and data level everything is measured in satoshi units, well actually its starts as binary then converted to integer, then converted to btc
NOT THE OTHER WAY ROUND

todays 6.25 btc is actually 625000000
in binary 100101010000001011111001000000

next halfs 3.125 btc is actually 312500000
in binary 10010101000000101111100100000

the halving is removing 1 binary bit from the end
so
1001010100000010111110010000 = 156250000
100101010000001011111001000 = 78125000
10010101000000101111100100 = 39062500
1001010100000010111110010 = 19531250
100101010000001011111001 = 9765625
10010101000000101111100 = 4882812
1001010100000010111110 = 2441406
100101010000001011111 = 1220703
10010101000000101111 = 610351

as you can see 9765625->4882812  does not have 2.5
as you can see 1220703->610351  does not have 1.5

thanks a lot. that is a very clear and logical answer. So indeed we will never have 21 million coins that way.

You can just ignore the nonsense from Charles-Tim, franky1, and hugeblack.  Clearly they didn't pay attention to what you were actually asking (although at least franky1 did point out how at a technical level the "halving" actually works and why the extra half a satoshi simply disappears).

As o_e_l_e_o has pointed out, and you have come to realize, it was never intended that there would be exactly 21,000,000 Bitcoins.  The more accurate phrase would be that there will never be more than 21,000,000 bitcoins, since no matter what the number is it will always be less than that.  There have on occasion been miners (or mining pools) that have failed to create the maximum allowed block reward in their blocks.  Therefore, the actual total will end up being less than the 20,999,999.9769 Bitcoins that were theoretically possible when Bitcoin first started.

The first block reward doesn't take part in the 21M rule as it doesn't take part in the first epoch. First halving occurred in block 210,000 and since the system halves the reward every 210,000 blocks, then genesis (AKA block 0) is excluded.

That's probably why Satoshi configured those coins as unspendable.

I'm not sure if coinmarketcap.com has had the correct supply: 18,836,593 BTC (as of block 703,865)

210000*50 + 210000*25 + 210000*12.5 + 73,865*6.25 = 18,836,656.25 BTC.

These 63.25 BTC aren't counted and are, most probably, provably burnt.

How can it break it?

Once we came into this agreement of four more decimals, the code would change, but Bitcoin wouldn't break. Satoshi (sat) wouldn't be the base unit (as it is in the source code), but another one which is a thousand times less than that. Transactions signed after this change could have a prefix which would indicate that the user signs based on the new milli-toshi base unit. Any transaction's amount which was signed without this prefix, like the ones before the change, would be considered as a thousand times greater.

the binary rule.. the real base rule.. the binary measure will be broken
but you think its ok because you can hide the break using user interface level code to hide the break with 'prefixes'

you are clamoring onto ideals that the hard rule is actually the soft rule of the user visual of btc.. without realising or admitting that the real rule is binary.

changing the binary of 6.26btc means that old nodes see the new measure as being more then 6.25 btc under their rules.
and new nodes see the old 6.25btc as being less than 6.25btc if some cludgy user prefix code is not added to hide the change

and yes it requires alot of cludgy code to translate and hope to not break the binary rule by having cludgy math express numbers different at human interface level

if you think you can have cludgy user interface code to hide the rule break than you are pathetic.
the other pathetic thing is how you defend such an act of editing bitcoins hard rule to be alt-net compatible. when these silly altnets you want compatibility with have no consensus nor longevity nor loyalty. and so it should be these altnets that should have the cludge to be bitcoin compatible

if other networks cant work with bitcoin. then other networks should not continue to operate as they have failed. if other networks want compatibility.. they should change. without breaking bitcoin to make other networks succeed

as always this question of "what if in 2140" and "what about 9th decimal" has been asked many many times. and so instead of just the obvious answer.. i go one step further and actually include details of CURRENT present tense issues that impact such issues of the topic, in the future.

the simple question has now been asked and answered. and now its for further lessons about more indetail stuff.. shame danny does not want people talking about the finer details of bitcoin. seems he does not want people to learn about the indepth goings on of bitcoin.. wonder why

but danny wants me ignored because i talk about issues that are in plan/discussion of change in the present and that affect the future of this topics question which he does not want people reading

but hey. this is a discussion forum about bitcoin not a "shut up about bitcoin and promote altnet" so i will continue discussing things that affect bitcoin

yep my comments change your highlighted
20,999,999.97690000 which should be achieved in 2140
to actually be
20,999,999.99999570 btc (user interface cludgy math total) by 2192
or technically
2,099,999,997,690,000 minimal sharable units changing to
20,999,999,999,995,700,000 minimal sharable units by 2192

meaning the topic creators question has more then one answer and result to his question

oh and the funny part..
discussing things dev want to be kept a secret(by ignoring/banning user discussion of such) is not 'fear mongering' its actually called transparency. and informing. because some people actually want to know about things being proposed that can break and change bitcoin.

pretending it can cause fear must be an admission by danny that what has been discussed is something people would fear. meaning danny and his ilk should not be secretly pushing to proceed in letting such a thing happen.
but if what was discussed had no impact and affect. then its not fear mongering. (so why call it fear mongering?)

so danny do you really approve of the change to the binary values and the need for cludgy code and math to translate and convert at user level and have alot more bug fix code to prevent back-forward compatibility issues just to get your altnet to work with bitcoin?
or
have i got you wrong and you want the binary measure to stay as is and not have "extra decimals" at user interface level, meaning you want to keep bitcoin values as is?

if its the second i apologise as pigeonholing you into the certain group. if its the first. then people need to know this stuff

That sounds like a reasonable number, but you're right it probably is a bit more.

In addition to the two links provided by o_e_l_e_o above, a similar review of spendable Bitcoins was done back in 2014. You may want to look through that thread to see what was found at that time and how the user went about it: