Here's vitalik's response.

"So here is a thought. Currently, because of the PoW ice age, the block time is scheduled to start increasing in mid-2017, and past around 2019 the increase is going to grow exponentially. The mining reward does NOT increase proportionately. Hence, there is already an exponential slowdown in the growth of the ETH supply built into the protocol; my script shows:

Block 3000000, approx ETH supply 87962556, time '2017-01-16 00:38:33.067775' blocktime 14.86 Block 3500000, approx ETH supply 90612556, time '2017-04-11 18:09:34.273529' blocktime 15.27 Block 4000000, approx ETH supply 93262556, time '2017-08-15 18:20:24.642729' blocktime 30.01 Block 4500000, approx ETH supply 95912556, time '2018-11-03 05:55:48.912370' blocktime 136.71 Block 5000000, approx ETH supply 98562556, time '2025-10-02 11:47:30.658317' blocktime 835.81 Block 5500000, approx ETH supply 101212556, time '2128-03-20 09:14:16.483692' blocktime 17183.83 Block 6000000, approx ETH supply 103862556, time '5189-09-26 20:57:59.367004' blocktime 520901.19

Hence, in the foreseeable future, the supply will not go far above 100 million.

PoS is likely to lead to quite low issuance rates; I am not comfortable promising zero, but if it is not much less than the current PoW then there is little point in making the switch in any case. If the community wishes to, while PoW is in play, it's possible to agree that any delay to the ice age bomb should also respect this general ETH supply growth curve, so in a situation where at time X if current ethereum would have a 75s block time, the ice age patch would set the block time to 15s and the block reward to 1 ETH (and adjust uncle/nephew rewards proportionately).

That said, issuance is a key economic parameter and I personally don't feel I or the foundation or client developers have the authority to dictate this; perhaps it's worth some kind of vote."

Btc actually has higher inflation I believe for another 10 yrs.