One of the biggest problems with rocket boosters is that you cram them full of amazing materials and engineering, and then ditch them after the first trip. It'd be like buying a Porsche 918, driving it to the coast, and then jumping out of it just before your let it drive off a pier. It's sort of crazy. SpaceX thinks they have the answer.
The idea of a recoverable booster has been around as long as engineer and accountants have felt a pain in their abdomens as they watch rocket boosters plummet into the sea. The Space Transportation System, which most of us know as the Space Shuttle program, did manage a way to recover the pair of solid-fuel rocket boosters, which returned to Earth on parachutes, were recovered from the ocean, and refurbished. There's some other launch vehicles that do similar things for their strap-on boosters.
The Falcon9 is the first booster to allow for recovery without the use of parachutes. The Falcon9 can restart its main engines to slow its descent, gimbal (that is, change the direction of the engine's thrust) the engines to keep everything nice and stable, and then eventually land vertically, on a set of extendible landing legs. Just like we've seen in so many cartoon and pulp sci-fi book covers.
That video there, despite the ice on the camera lens degrading the image quality, shows the process in action. The rocket is designed to tip horizontal when in water, which makes sense, though in this test it seems the impact onto the water caused some issues with the structural integrity of the rocket itself, but SpaceX thinks they'll have that licked for next time.
Last year they tested the fundamental system on a suborbital Grasshoper rocket launch, and this video gives a much better look at how it all works:
If SpaceX can reliably recover their boosters, it means a pretty significant shift in the cost of getting things into orbit — potentially the price could fall to as low as $500/lb, which is about 1/20th of what it costs now. That also means that turnaround time would be dramatically better, since the booster would only need to be checked over and refueled before it's ready to go again. Which could mean more launches, more volume, less cost.
Plus, think how cool it'll look at the spaceport with all these rockets gently dropping down to land. I'm picturing them coming down like rain, which won't, of course, happen, but try and stop me.