What's the point of loading that into FPGA, when it is basically an abandoned platform? They are not going to support any of this open-source nonsense as soon as they get closed designs out.
That's your opinion, to which you are entitled, though I'd be more comfortable if you added an "I work on ARM stuff for Atmel" disclaimer.
Your opinion is contrary to what the SiFive founders, several of whom are also still academics at Berkeley, say. They'll lose face more by changing the plan than they will by failing commercially.
A major advantage of open source with a free license, of course, is that once it is released it is out there forever, regardless of whether the originator ceases to exist, or makes later improvements closed (as the license entitles anyone to do, not only SiFive).
Note that the thing they are offering for a $275k fee is primarily "free" support, along with customization and some extra peripheral busses.
My own disclaimer: I currently work at Samsung R&D Russia on a project for "Device Solutions", the chipmaking division of Samsung, an ARM architectural licencee that designs some of its own ARM cores and SoCs (e.g. Exynos). However I'm working on another kind of chip and have no knowledge of any future ARM, RISC-V or other CPU plans.