This is a very puzzling claim. The overall yield of EMIB cannot exceed the yield of the silicon bridge embedding process in the ABF substrate.
I have seen numerous reports on EMIB yields over the years, but most have provided only fragmented information. I suspect that very few people actually have an accurate understanding of the yields at each individual process step.
To the best of my knowledge, the yield of conventional EMIB (non-EMIB-T) is significantly higher than the figures cited in this report.