There was also another major warning in his speech. The government taking over and driving the civilian research and development and the money pit that was going to be. I’d guess that was rooted in the massive monies spent on the Manhattan Project. And then in the term following Ike’s we got the space race from Kennedy. Both highly useful and needed research efforts and both massive spending efforts bending civilian scientific professionals to the government agenda and paid for with government funding.
In both cases he called neither a boogeyman. He pointed out they are in fact needed but don’t become beholding to them.
Some of the cost problem has to do with trying to make huge jumps in military hardware. A couple of examples are the extremely expensive F-22 and F-35. Capable probably, but really TBD. I'll give an example from my background in developing a system for use in nuclear power plants. A Loose Part Monitoring System (LPMS) is used to detect the presence of and hopefully determine the damage potential of anything that has broken loose within the reactor coolant system. It consists of accelerometers at strategic points on the system, the necessary instrumentation, and generally some diagnostic tools. The early systems were analog, the system if it sensed anything similar to an impact set off an alarm, and with most systems an operator could listen to an audio monitor for impacting. Most systems added tape recorders - problematic when most recorders didn't do more than four channels and an LPMS had between 10 and 20 channels - largely depending on the manufacturer.
Within a few years, digital advances gave the potential for better discrimination ... fewer false alarms, and the potential to give the order in which channels went into alarm - the ability to determine where the sensed noise was loudest and time differential to triangulate the source. Recording was still a huge problem. If there was one impact like event, it was gone before the tape recorder ever got started, and the ability to digitize and store the data was well into the future. In my last job digital technology began to catch up with the needs. We could store digitized signals (roughly audio range - up to 20KHZ) - just not much at a time. There was still a problem with recording time, and most of all inability to record the initiating event. One day it clicked for me - back to time delay technology for moving target displays in radar. Why not digitally delay each output for some set period of time like 5-10 seconds and record the delayed signal, based on an alarm from the un-delayed signals. Worked like a charm the system alarmed, started the tape recorders which were up to speed when the delayed impact reached them. Now with far better processing power and tons of storage, you don't even need the analog recorders.
The point is sometimes what you want or need to do isn't technically feasible. Instead of trying to deliver a hugely complex and expensive system before the technology is really there. Why not build a modular basic system that can be upgraded as technology allows or becomes available. Just recently guys in an F-35 came up with a new data recording system to replace an huge and insanely expensive system $230K vs $25M each and reasonable in size and weight. This is an example of something driven from the bottom up rather than the top down; an example of something we can do in the future but not properly now; and an example of people who really understand the issue being the ones to develop the solution. Let's build the basic plane and add some of the bells and whistles as modular addons instead impacting the system with huge up front costs for things that might not be the best way to go, aren't fully developed, and take advantage of the fact that not all technologies follow the same time development path.
Right now the AF is trying to dump the F-22 because several of the systems are outdated while the plane itself is apparently still extremely competent. What if the plane was considered one item and target acquisition and weapons systems were separate items - the ability to remove one APQ whatever and install a newer one. Reasonably updatable electronics in the same airframe - it's done now, but why not planned to do it more efficiently. For the most part electronics are decreasing in size rather than requiring more space. We get all choked up on trying to do everything at once rather than a bit at a time. Specify and build the basic prototype; then figure out if you've got something useful and what you can then do with it. And for once in the present time, build something to do one thing right at reasonable cost and quit trying to build flying Swiss Army knives.