I wonder how much it costs them in the long run; the initial number mentioned seemed cheap enough (equivalent to 1-2 developers for a year - and I don't believe 1-2 developers could rewrite it in a year, for various reasons), but if it's a 1:1 "transpilation" to unsafe Rust... was it worth it? Will it be worth it in the long run?

But this is the longer running discourse about AI code's cost / benefit. I keep nagging my work's AI community to show me the numbers, so far nobody has been able to show an increase in productivity, quality, customer satisfaction, or company profits.

it said in the original article on the bun rewrite ~165K.

For that money they could hire engineers to do it without having so many unsafe blocks.

The only gain is speed.

I saw an estimate of around 160k $ somewhere in HN comments about the rewrite.