This is yet to be proven. Is the bitcoin network really more costly than existing payment systems? Can the security benefits, instantaneous transmission, and other features be replaced by less costly alternatives?
As the mining reward rapidly dwindles relative to the cost of electricity, there is a huge incentive to reduce the electricity consumption while maintaining the security of the network. At what point does this reach an equilibrium depends largely on the market value of bitcoin. So it should be a somewhat fixed cost in proportion to the adoption of bitcoin, just like the fixed costs of any currency (e.g. minting coins, mining gold, etc.).
Either Bitcoin is just a bubble, and it'll stop, or it's actually useful, and not just "imaginary points". Either way, it'll probably be decided way before it has any real impact on the global energy consumption.
At present we're in a 'gold rush' phase, and the rewards from mining result with a large amount of resources being spent on it.
Mining rewards will diminish over time so it will no longer be feasible to mine as much as in the 'gold rush' phase. So the carbon footprint will not be growing all the time.
In fact, there's a race to make the current hardware to become more power efficient. If you live in a sunny place, you could probably pick up 240 Watt solar panels for around $300, a connect a Raspberry pi @ 3.5 watts, then a 10 GH/s Erupter blade @ 75 watts.
Is there a better way of doing it? There needs to be a way to distribute the first bitcoins fairly and anonymously, and it's also a great way to get people to start using the currency. It would be ideal if the bitcoin mining stopped sooner than it does, but it seems to have been enough to jump start the currency to where it is now.
As the mining reward rapidly dwindles relative to the cost of electricity, there is a huge incentive to reduce the electricity consumption while maintaining the security of the network. At what point does this reach an equilibrium depends largely on the market value of bitcoin. So it should be a somewhat fixed cost in proportion to the adoption of bitcoin, just like the fixed costs of any currency (e.g. minting coins, mining gold, etc.).