Considering I have never even HEARD of another model cassette breaking in this fashion from ANY of the Big 3, it means a lot of DA 9000 cassettes are breaking.
Despite me posting about the (fairly widespread) failures due to a poorly designed XT-9 cassette spider on this very thread?
Maybe the DA MK2 cassette should use a 5 point mount, rather than 4. (that was shimanos fix for the XT failures.)
If so, could this be getting in around the rivets and causing delamination/weakness?
It's carbon fibre reinforced plastic, rather than a cut sheet. So nothing to actually delaminate.
Actually looking at the design, could it simply be the difference between riveting a DA/titanium sprocket (quite a springy material) to the carbon reinforced plastic and riveting the Ultegra/steel sprocket (~40% higher youngs modulus) to the same/similar carrier? The springier ti sprocket moves around a lot more, allowing the plastic to fatigue? The steel one, being stiffer, spreads the load across more of the rivets. Once you start looking at tolerances on the riveting process (end load etc) you might start to see those cassettes with the poorest stack up of tolerances/thicknesses/rivet end load failing prematurely. As they move around the most.