1) If there are some widely used open hardware devices or open standards. The fact that FPGAs are proprietary designs and can't be copied is a significant barrier. If there was an analogue of ARM in the FPGA world that could open up open source dev tools too.

2) DIY FPGA. This one is almost certainly much further down the road than any of the other options. But if fab and design tools of the chips themselves become a commodity then there will be open source dev tools. It used to be pretty difficult to make a custom PCB, but now with PCB as a service open tools are seeing more use. I think the reason this is so far down the line is because fpga offers cutting edge performance for specialized applications. If you wait several generations or are ok with reduced performance then you can stick with generic CPU/GPU. With Moore's law running down there may be more and more things that aren't possible without fpga. And I don't think it would be until then that open source dev tools follow more diy fpga.

https://symbiflow.github.io/

There is an open FPGA hardware implementation:

https://github.com/haojunliu/OpenFPGA https://www.eecs.berkeley.edu/Pubs/TechRpts/2014/EECS-2014-4...

More resources:

https://wiki.debian.org/FPGA

For my comment I was thinking about HDL simulators, which are very expensive/proprietary tools (vendors like Synopsys, Cadence, Mentor Graphics), with a lot of vendor lock-in, but yet are based on standard/open languages: Verilog/System Verilog. I know there are some open-source simulators that have been around for a while, but they aren't widely used, and the big 3 EDA vendors still seem to have a strangle hold on the HDL simulator market.

Simulation is generally more important for ASIC hardware than FPGA hardware, since you can quickly iterate on an FPGA design with logic/design fixes, whereas with ASICs design iteration can be prohibitively expensive and time consuming.