So how does that architectural rewrite fundamentally change what developers can realistically build and execute at scale?

Yeah, so Qiskit 1 was the start of this stable semantic versioning, right? So researchers can be confident that the code and experiments they wrote in Qiskit 1 would be compatible across future Qiskit 1 releases.

And similar to now, we have Qiskit 2.x releases as well.

When we think about the start of performance, back in, I think, 2020, the Qiskit transpiler really struggled to scale up for a system we had at the time.

It was a 50-some qubit experiment.

Transpilation, for those who don't know, is a core part of every experiment.

You have circuits that you write, and to prepare them for execution on a quantum computer, you need to set up the right layout, synthesis passes, a whole bunch of different techniques to make sure that This abstract circuit you wrote is ready to run on hardware.

Sanket, do you want to take that first?