If you look at HFT as an industry or a business, as opposed to valuable speculation, then it certainly has a life cycle. I imagine the HFT companies of today are like the dot coms of 1999. The question is then become show does HFT end?
I'm referring primarily too predatory HFT algorithms which use latency advantage to take sniff out institutional orders.
I will post some hypothesis below
Technological advances from the buy side.
If you imagine that it takes say 200ms for an order to hit the exchange and a firm wants to buy. The HFT algorithm detects the buying and buys above the level or sells below it. It is easy to see that even with a 200ms delay that if 2 orders are sent near simultaneously that is a large sell order and a large buy order that the HFT algorithm will not be able to respond fast enough because there will be near zero lag between orders. I've read some firms are offering technology like this already.
Exchange rules & Regulation
It would be easy for the exchanges to end or curtail HFT by simply requiring all posted offers be valid for a minimum set amount of time. I.e if you post an order then it must be available for a minimum of 500ms before canceled. I've read about messaging limits but that makes the exchange money which isn't desirable for HFT deaths spiral.
More and more HFT algorithms competing for the same profits.
The risks the HFT traders take may not be fully understood by the participants.
Look forward to ideas/thoughts.
HFT is the new way of trading in today world. It is in the beginning stage. The only way HFT ends is when electronic trading ends. Otherwise, this is the robot trading era.