Single buyer does not need to mean no innovation.
The NIH spends a lot of money on basic research over many researchers, and it is seemingly doing well.
A bit of background which may help explain how innovation is done in DoD.
The congress appropriates research and development money for DoD to pursue innovation, that is stay ahead of adversaries with new mission capacity and do the missions we already do cheaper or better. R&D money is broken up into at least 6 sequential ‘buckets’. The first bucket is spent on things that advance basic knowledge, the next bucket may develop a function that the knowledge or ideas make possible to do, like fly around invisible to radar, then the next bucket develops a prototype, then you may expand the prototype to work in some environmental conditions. Do some tests and see what you can expect to do in terms of performance and cost. Some technology money goes to developing manufacturing and repair capacity as well.
Once you have gotten the innovation to the point where you have run some tests on a physical and SW enabled model you write a development specification for industry to use to take the ideas as somewhat mature technology and make a product that could do the job at a reasonably estimated cost to make and operate. Here is where the conspiracy of hope and need to keep industry busy breaks down the logical process.
Developing and testing the product is paid for with the 5th and 6th buckets of the R&D appropriation.
When the tests pass (conspiracy of hope is a problem here at DoD) then you get production appropriation money to make the product and test it with the operator in their hands in the real world.
If that test is “good”, almost always because the congress now sees profits, the system goes into production and is fielded.
NIH already pays for at least the first couple of innovative R&D steps.
Like in early stage DoD R&D, this money goes to basic research at universities and some private labs who writes grants to define the science and possible benefits of learning about the object studied.
The DoD porcess is good if run effectively and if the conspiracy of hope were ended and bad technical ventures killed when they should be.