They [1] did not make not a quantum computer. They made a BosonSampler, which is a different way of demonstrating quantum supremacy. A BosonSampler is not a uni

They [1] did not make not a quantum computer. They made a BosonSampler, which is a different way of demonstrating quantum supremacy. A BosonSampler is not a universal quantum computer because it's circuit depth and/or width would group exponentially in the size of BQP [2] problems. Very much like Google's chip, this machine is only good for solving one useless task: sampling from a probability distribution formed from the permanants of the nxn submatrices of a mxm unitary matrix, where m>n. This task is hard for classical computers but easy for quantum devices because you can just make this chip and each run returns a sample from the distribution. The initial estimate threshold for showing quantum supremacy via a BosonSampler was n=30 photons in m=100 modes, but this might have increased recently as classical simulations methods have improved. These guys have 20 photons in 60 modes, so they are getting close. [1] https://arxiv.org/abs/1910.09930 https://arxiv.org/abs/1910.09930 [2] set of problems solvable by a universal quantum computer efficiently.