Quantum computers can offer dramatic speed-ups over their classical counterparts for certain problems. However, noise remains the biggest impediment to realizing the full potential of quantum computing. While the theory of quantum error correction offers a solution to this challenge, a large scale realization of fault tolerance seems currently inaccessible. What can one hope to do then, with existing noisy processors? In this talk, I will present experiments that produce reliable expectation values from a noisy 127 qubit processor, at a scale that is well beyond brute-force classical computation. We argue that this represents evidence for the utility of quantum computing in a pre-fault-tolerant era. I will also discuss recent classical benchmarking of our experiments beyond exact verification.