In a way, centralization seems unavoidable. The governments that fund research have themselves become far more centralized, so perhaps science has been pulled along in the process. But even without that prevailing wind, science would, I think, head in the very same direction.
A young discipline is bound to move first through the data it can gather most easily. And as it does, it also defines more exactly what it must measure to test its theories. As the low-hanging fruit vanish, and the most precious of fruits are spotted high above, bigger investments in harvesting equipment become necessary. Centralization is a way to extend scientists’ reach.
I think I would draw a distinction between "Big Science" versus "Big Facility". While it is true that in many instances, "Big Facility" = "Big Science" such as the Tevatron and the LHC, there are instances where one could argue that they are not automatically the same thing. I would say that a synchrotron center is a "Big Facility" but not "Big Science". The science being done at such facility comes from such a large and varied subject area. Such a facility could easily accommodate fields as diverse ranging from pharmaceutical to biochemistry to material science. Many of these sciences are not what one would consider as "Big Sciences", i.e. they do not normally operate out of such huge facility. Material science/condensed matter, for example, often operates out of research equipments that can fit into a standard-sized laboratory space. The availability of a synchrotron facility simply adds another "probe" and capabilities that are convenient and useful as another experimental avenue.
So while the LHC and Tevatron and ITER are all "Big Science", synchrotron centers and neutron facilities etc. are merely "Big Facilities", not necessarily "Big Science".