1) The companies pursuing this, SpaceX, Blue Orgin, etc., can't be the first to think of it. If the first stage accounts for 75% of the cost of a launch, as one article I read says, I'm sure many have considered, going back to the first launches decades ago, how to reuse it.
2) The technology to land rockets vertically has existed for a long time, going back to the lunar lander at least.
Additionally, Blue Origin and SpaceX have different goals from traditional aerospace firms that requires them to drive costs down below what a government would require. Blue Origin want to offer rides to space to tourists; for that, their costs need to be low enough for a person to buy a ticket. SpaceX wants to land their rockets on Mars, and then take off again. Being able to land the first stage of the Falcon 9 allows them to perfect the landing and reuse of rockets on Earth, before taking that knowledge to Mars.
Usually changes like this result from a technical or economic development.
And they have a drive, too. Mars is not a marketing story, Elon seems to really believe in doing it, and a lot of people at SpaceX buy into this vision.
Landing the Lunar Lander on the moon is completely different to landing a tall, thin high-centre of gravity rocket full of explosive fuel, using AI because there's no crew onboard.
Or did you mean something else?
I'm pretty sure the first vertical landing of a rocket was only achieved in the last few years.
This development trajectory leads to a capable launcher, but one that is wholly unsuited for reuse. Reusing the first stage is now extraordinarily difficult, extraordinarily costly, and has little advantage (since most of the cost is in the upper stage anyway). You see the effects of this design pattern in most expendable launchers. Both the Delta IV and Atlas V boosters use a single engine in the first stage, and both use an RL-10 LOX/LH2 engine for the 2nd stage, an extremely expensive engine on its own, let alone when it's integrated into a high-tech stage (such as Lockheed-Martin's legendary Centaur line). The Ariane 5 has gone through 5 different generations of upper stage advancements, and it too has only one engine on its first stage. India's GSLV uses a solid first stage with hypergolic boosters and second stage and a LOX/LH2 3rd stage.
Meanwhile, there hasn't been much pressure on rocket designers to be hard headedly pragmatic. Often times winning government contracts or even private investment meant convincing people of your beyond state-of-the-art designs that would leap frog the competition, rather than rely on common sense engineering and iterative advancements. That's how we got the Shuttle (which was supposed to be about 2 orders of magnitude cheaper to operate than it turned out to be), the X-33/VentureStar program, the NASP/X-30 program, Rotary Rocket, Pioneer Rocketplane, and all the other programs and startups that were cancelled when they ran out of money, ran into engineering problems, or both.
The sensible thing is to try reusing some of the rocket and make the most of that from a cost perspective. Something that saw a bit of attention with the old DC-X program over two decades ago. One problem with that program was that it envisioned a new launcher that was mostly an SSTO, and aimed at building up from small scale rather than building toward reuse from full orbital scale.
The genius of Musk and SpaceX has several facets:
1) Begin with expendable, but design with reuse in mind (high throttle ability in the first stage, most of the costs in the first stage), and work towards there iteratively. That way you get experience in a lot of the most important bits of the business: building orbital rockets, building rocket engines, building spacecraft, launching rockets. And you get cold, hard cash flowing into the company from launches. Cash that keeps the company in business, and finances further innovations.
2) Build reuse not as a different vehicle but as a "feature flag" on a common design. The difference between a reusable launcher and an expendable launcher for SpaceX is a mode selection, not an architectural difference. This means that you get operational testing of the vast majority of your flight hardware for your vehicle with every single expendable launch. This is a well-known risk mitigation strategy in software and other industries, and here it works marvelously. Prior to today's launch SpaceX had 14 previous launches of substantially the same hardware (counting only the v1.1 version).
3) Opportunistically use the first stage after stage sep. for experimenting with boost back and landing. From the very first v1.1 launch SpaceX has been aggressive at testing aspects of the landing and reuse flight mode, equipping stages with landing gear in many cases. This is a fantastic cost saving method as it takes what would be trash and transforms it into test vehicles being operated in typical flight profiles. This takes advantage of hundreds of millions of dollars of flight hardware to conduct an R&D test program on the cheap.
Combine all of that together and you end up with a way to get from A (expendable rockets) to B (reusable rockets) without having to take on a ton of risk, without having to spend insane amounts of money, without having to design and operationally prove a brand new reusable vehicle design, and to build a revenue earning and self-sustaining business before being able to finish reusable vehicle development. It took a significant amount of foresight and savvy to be able to see that route and to execute on it.
And yes, knowing what we know now it would have been easy to do this sort of stuff in the past, more so if we wanted to just throw money at it. But it wasn't obvious until being proven.
The fact that the Falcon 9 is deliberately designed for less efficiency with 9 engines on the first stage, but still ends up cheaper than the alternatives is even more impressive given that insight.