What most people don't grasp (including the author of the article in the OP) is that the amount of electricity going into mining BTC is correlated to the amount of new bitcoins being mined - not to the total market cap of bitcoin. In other words, if $1 million of new bitcoins in being mined each day, than it makes sense that just under $1 million of bitcoin mining expenses (ASICs, technicians, warehouse rental, electricity and so on) will be expended in the effort to mine bitcoins. But if that drops to $500,000/day in new bitcoins, then the amount expended on mining will likewise drop in half.
Halving the reward = halving the security of Bitcoin network. Unless, of course, miners manage to compensate for the halving block reward with higher fees. In which case the total reward/power consumption remains the same.
So what does this mean? It means that with every halving the investment in bitcoin mining - including electricity expenses - is going to drop by half, all else being equal.
Correct, if BTC price does not go up, Bitcoin network will consume less electricity with each halving. Of course, Bitcoin will also become half as secure with each halving.
In 2020 we will be having another halving, so after that point the rate of bitcoins being mined will be 1/4 what it is pre-2016 halving. The only way bitcoin will consume as much electricity as Denmark is if the value of each bitcoin rises to stratospheric levels, which will probably only happen if there is a wholesale collapse of fiat currency systems.
By "stratospheric", you, of course, mean 2 * 2 * $450 = $1,800. Or just $600 more than it was worth in 2013.
Stratospheric indeed