What I'd really like to see is another manufacturer launching an FPGA that uses the same bitstream format (and hence the same general architecture) as the iCE40 or ECP5 series.
Then we'd have a proper open platform! Remember the "good old days" when chips were second-sourced? (And third- and fourth- sourced for the really popular ones: Intel's 8086/8088 come to mind here.)
NDAs and Patent issues are number one in this. It's the same reason AMD can't open source their platform security processor and why it took so long for AMD to start being able to contribute to open source Linux drivers.
Going through and open sourcing any part of a modern piece of silicon or the software interfacing with it is a miserable process. Basically everything has to get arduously churned through legal to verify that it doesn't expose the company to any risks.
Unfortunately that open sourcing process is largely thankless work and as a result, companies either never make the leap or end up dissuaded during the multi-year process of clearing the "to be open-sourced" code through legal.
Thanks. Your last paragraph is the main argument from my point of view.
Without corresponding court cases, the assessment by the legal services is only based on best effort anyway. They therefore do not want to let themselves out on the branches and tend to come to conservative assessments. And they already have enough work to do and the importance for the firm is also limited, which explains the long processing time well.
Then we'd have a proper open platform! Remember the "good old days" when chips were second-sourced? (And third- and fourth- sourced for the really popular ones: Intel's 8086/8088 come to mind here.)