I don't think this argument is sound. In your EV calculation you're including the expected deaths over the thousand year period but excluding the expected lives over that same period. There’s an asymmetry in this comparison. 

Also, I don’t see how x number deaths of a given species could be worse than the extinction of that species. The way I see it the first choice is save k lives over a thousand years, but the second choice is save k less lives over the same period and loose all future lives after that, forever.

The government should defend against the second case.

In the summary you mention that "Skepticism of formal philosophy is not enough". I’m new to the forum, could you (or anyone else) clarify what is meant by formal philosophy? Is the statement equivalent to just saying "Skepticism of philosophy is not enough" or "Skepticism of philosophical reasoning is not enough"? 

Also, in the section "Increasing Animal Welfare Funding would Reduce OP’s Influence on Philanthropists" you make a comparison of AI x-risk and FAM. While AI x-risk reduction is also a niche cause area, I think you underestimate how niche FAW is relative to AI x-risk. The potential alienating risk from significant allocation to x-risk isn’t the same as that of FAW since AI x-risk is still largely a story about the impact this would have on humans and their societies.

I’m not saying this is the correct view but the one that would be generally held by most potential funders. 

In general the utilitarian case for your main points seem strong. Great post.