It’s a sliding scale in the end. You can have a fusion system that barely breaks even, or only slightly breaks even, and then is still more expensive than traditional energy sources after factoring in fixed costs and maintenance.
Even then, are there no sales possible? I'd think that a borderline fusion reactor would be commercially viable if only for research. You could probably sell a dozen of those around the world. But unless something really weird's going on with the physics, a borderline system is just the mark 1, and everyone expects mark 2 to be more efficient.
For example, a DT fusion reactor based on ITER would be two orders of magnitude larger (in volume and mass) than a fission reactor of the same power output. It might "work" in some technical sense, but be completely economically impractical.