A Boltzmann machine, an Ising model and a probabilistic-bit fabric are one object under three names. Hold such a system at a temperature and leave it alone, and the fraction of time it spends in each configuration converges to the exponential of minus its energy over temperature, divided by the sum. That is the Boltzmann distribution, and it is the same object the sampler in lesson five was trying to reach by running a chain in software. On a von Neumann machine you simulate the settling. On a fabric built from the right devices you do not simulate it, because the devices settle. This is why the argument is not that the new hardware is faster. It is that the operation you were paying for is what the device does when you stop forcing it.