Decentralized exchangers will not only be important in the future, but also centralized ones.
Decentralized exchangers have great advantages like high privacy and no influence of authorities, but they are also expensive. Without managing the trading platform, trades must be made directly with transfer fees for each trade and you have to manage the security yourself or find interactive smart contracts.
Centralised exchanges are definitely still in the picture for the immediate and foreseeable future. The fact that they've been around for years, and will cater to the beginning needs of newcomers - along with the fact that a vast majority of people will actually still prefer to choose centralized ones for the perceived benefits of a caretaker/arbitrator, or in the case of regulated ones like in Japan - insurance. Most people can't even trust themselves to remember their email passwords, much less private keys or seeds that aren't even in their native language.
But the future also belongs to decentralised exchanges, and we already have some making small Waves (forgive the pun) in the industry. I try to use them as much as possible, despite their painful shortcomings (low volume, thin orderbooks). Risking a shill remark, my eye is on a project called NVO but there are many others emerging with beta platforms which should go public this year. 2016 was when the talk arose, this year could possibly see them take off.
