If you look at most published articles, or those on the preprint server, you find the vast majority of articles have more than one authors. Are collaborations needed to be productive in physics? In other fields like math, single author articles are the norm.
Nobody seems willing to try and answer this soft question, so I will have a go.
The answer is yes. Experimental High Energy Physics is the worst, with 3000 people signing the papers in the LHC, even most theoretical papers have at least 2 names.
The basic reason, imo, is that physics has become a discipline where in order to keep everything in your head you have to have special talents, like an eidetic memory. A middle of the road physicist has to rely on interactions outside his/her head, and what is better than a colleague to thresh out a problem or a proposition. It is a peer review right there and then, so useless ideas can be discarded fast and new ones explored. Efficiency goes up for success in publications. This holds for theorists and disciplines where one could think that one person could carry off the whole paper. It is more productive to discuss the snags with somebody else, before submitting a publication.
In the case of experimental HEP the added burden is the enormous complexity of the experiments that need teams of people coordinated to build detectors and innumerable shifts to get at the data. It is group work, like building a cathedral.
There are two things at work here: First, physics has advanced to the point where all the interesting problems are sufficiently complex that they almost require multiple people working on them with different areas of expertise. It's not impossible to do publishable physics research as a single individual-- it's more common in theory than experiment, but you do occasionally see single-author experimental papers. (Not in high-energy physics, but high-energy physics isn't the whole of physics by any stretch.)
The second factor is that modern physics is relatively generous about co-authorship. There are other disciplines where you have to have been far more involved in every aspect of the work to get author credit than you do in physics, where contributing to a sub-part of the data analysis will often get authorship credit.
This hasn't always been the case. If you look up some classic papers from the early 20th century, you will often find people thanked in the text for their invaluable contributions; contributions that, when described, make it sound like the person thanked really should've been listed as an author, by modern standards.