The technology wasn't ready.

Minecraft was groundbreaking in optimizing the handling of destructible maps. This was during a time where every other game were trying to do realistic graphics. Using blocks or voxels was considered silly because it couldn't ever look realistic. The game Delta Force used voxels for the landscape and even if it looked nice for that, it also showed that the use of voxels was not ideal for detailed realistic graphics.

Destructible landscape was used in 2d games like Worms, but that was basically just a bitmap image, so it didn't require much memory to keep track of the limited map size. Minecrafts map is theoretically infinite, and while some other games also had chunk loading, it was new that the game was able to save the precise state of all the unloaded chunks. The magic is that the game only saves changes that are made. The rest of the map is procedurally generated, so it doesn't have to save everything. It can just recalculate the map and apply the changes. No other game did this.

Gameplay wise, the survival game Stranded was close but it was limited by using ordinary polygon graphics, so while it had crafting and some mining (gathering materials), the map itself was not destructible. It also had a storyline so it wasn't as much of a sandbox game.

So basically, Minecraft combined a lot of different technologies to make it possible and it was only possible because the graphics were grossly simplified. It started as a hobby project and I doubt that any AAA publisher would have funded it at the time, because the idea of block voxels was so far out.