AMT... I don't recall saying that IMET handled mounting boards on heat sinks.
It explicitly says: "AMT was responsible for the mechanical designs, thermal designs (heatsinks), assembly, final test and QC. "
Also, apparently AMT is trying its best to distort history by finally posting some news in the past: https://advancedminers.com/category/amt-news/ that was not there previously
Specifically the last sentence of 1st paragraph.
It in turn brought a response about testing for more than a few min. Anywho that usually means that sinks were used that could not handle extended running. Not uncommon for limited full power go/no-go test rigs - but - ONLY after longer testing proves that it is good enough and then they were removed. Sadly once again, no time was allowed for proveout. Was this the case? (test-only rig)
Given the nature of these beasties personally I would do only full power testing after the boards were attached to their real-use heatsinks... At least the big'uns on the back. For short ter-testing the chips would run topless so to speak. A lot less trouble/expense than mounting to test sinks & then removing them/cleaning up just for the real heatsinks to attached later on... However, that is not what IMET said, they only talked about heatsinks for testing and nothing beyond that.
Back to the show.
Like any engineering enterprise, you got to validate some initial prototypes work out the issues then go towards mass production. The Bitmine and AMT boards were labeled revision 3, so there were at least 3 attempts to fix issues. In fact, the latest Bitmine board doesn't look anything like the previous AMT rev. 3 board. It has 10 chips instead of 8 chips.
Now if a company takes the risk of going for a full production run without adequate testing, that is a risk they take. However, that's the gamble you make when you claim delivery earlier than everyone else. The penalty for failure should fall on the company and not the customers.