Funding open source scientific software is a huge part of what the NIH does!
Unfortunately there isn't a requirement that the code developed under grants is released as open source, but there is a lot of movement to make it so. Furthermore, researchers have strong incentives to provide reusable software implementations of their research. (I know many people say the incentives run against this, but it is now clear that useful software gets more citations and scientists forget so at their own risk!)
I don't have a recent budget handy, but at least 10% of research funding is going to various forms of software development. I am happy to be shown wrong but 10% seems like a low estimate as literally all analysis is mediated by software. It's not like people crack out slide rules when they get a data set from a sequencer or mass spec. They go to GitHub and download some obscure domain-specific open source analysis package. Point being: a tiny administrative change could ensure that we direct $3B a year to open source scientific software in biomedicine! And further, it nearly looks this way already.
Unfortunately there isn't a requirement that the code developed under grants is released as open source, but there is a lot of movement to make it so. Furthermore, researchers have strong incentives to provide reusable software implementations of their research. (I know many people say the incentives run against this, but it is now clear that useful software gets more citations and scientists forget so at their own risk!)
I don't have a recent budget handy, but at least 10% of research funding is going to various forms of software development. I am happy to be shown wrong but 10% seems like a low estimate as literally all analysis is mediated by software. It's not like people crack out slide rules when they get a data set from a sequencer or mass spec. They go to GitHub and download some obscure domain-specific open source analysis package. Point being: a tiny administrative change could ensure that we direct $3B a year to open source scientific software in biomedicine! And further, it nearly looks this way already.