Sure they spent billions on new signals... But signals are really just lights in boxes. If I gave you a pile of raspberry pi's, some light bulbs, and a few months salary, I think you could do it.
The expense comes from the ancient methods they wanted to use. In the same way a handwritten book is much more expensive than using a printing press.
The new signals are trains communicating their exact location with each other and a centralized control system, and determining their speed based on the train in front of them. The same kind of vehicle-to-vehicle communication that doesn't really exist in personal automobiles today, despite billions of dollars in investigations. You can't use LIDAR or anything used in commercial AVs, because trains are much heavier and take longer distances to come to a complete stop, and line of sight is not sufficient.
If a satisfactorily reliable solution for such a problem could be engineered on the cheap, a private company offering it for much less would make a fortune off of all the railways chomping at the bit to save billions of dollars. The fact that no such solution exists anywhere leads me to believe that it isn't trivial.
The expense comes from the ancient methods they wanted to use. In the same way a handwritten book is much more expensive than using a printing press.