MPW
A Multi Project Wafer is used to reduce the cost of making an ASIC. By using a lot of different designs on the same wafer the Non Recurring Engineering (NRE) costs of making the mask set are shared between all the participants.

The Google shuttle is an example of an MPW.
Our MPW submissions
So far we have made 4 tapeouts to the Google MPWs:
MPW physical implementation
We sent John McMaster some chips to take a look at from MPW1. Here presented his findings at OpenTapeOut
And you can browse his microphotographed die here
Course feedback
The part I enjoyed the most was going back to the Verilog and seeing test benches pass. In particular the waveform viewer. At the latter parts of course you're simulating the entire Caravel system on chip with a RISCV core. Being able to drill down into everything inside that core, I can log the program counter, I can log all the address and data buses and you can just see in exquisite detail what the system is doing, and it's doing that because I programmed it. Being able to drill down into that detail was really fascinating.
Jonathan Pallant (digital course)