In my lifetime, there will be at least two major challenges to human society: climate change (of which you have hopefully heard) and peak oil (of which not enough people are talking). The first necessitates decreasing emissions, going green, reducing consumption, likely a reorganizing of human society. The second necessitates reworking the entire food chain (agriculture, processing, transportation), manufacturing of most goods, energy production, and again likely a reorganizing of human society. Let us be slightly optimistic. Let us assume the species will not destroy itself and that there will continue to be a high technology society. Let us also assume that due to the severe environmental constraints (stated above), energy supplies will be very restricted. There will be lots of competition for resources such as water and arable land, and also for green energy sources (hydropower, solar, wind, the usual suspects). There will be no oil or coal power plants that would provide nearly unlimited amounts of energy at very little cost. Nuclear fusion is still unusable, and will remain to be so (in large scale) for a couple of decades at least. (Nuclear fission will not remain a valid option once the world runs out of oil, as all mining vehicles run on diesel or gasoline -- and if you can't mine uranium, you can't run a fission reaction.) If you start doing a PhD in physics right now, which branch does it make most sense to specialize in? By that I mean which branch will still have jobs when there is no more cheap energy. As an example, the LHC can consume 180 MW at peak power (http://lhc-machine-outreach.web.cern.ch/lhc-machine-outreach/faq/lhc-energy-consumption.htm). You would need about 150-200 wind turbines in a good location or a large enough hydroelectric plant nearby to power just the LHC. In a resource and energy restricted world, this is not a sensible allocation of energy use. Another example is astrophysics. I could not find any figures off Google what the typical power consumption of an array of telescopes is (I would be interested if you have any!), but I would guesstimate it to be larger than a couple dozen contemporary households. In addition to that, computing, say, the collision of two large galaxies on a supercomputer may not be possible at all; IBM Blue Gene/Q, for instance, draws about 6 MW at peak operation. (http://www.hpcwire.com/features/Lawrence-Livermore-Prepares-for-20-Petaflop-Blue-GeneQ-38948594.html [Broken]) And that is considered a very energy efficient supercomputer by the way. I suppose my question can be rephrased: which branch of physics can do away with experiments with least energy use? HEP is out (in the form of mega experiments such as LHC, and that's where the remaining gaps in our knowledge are); most computational stuff is out; large telescopes may be out. What is left? Condensed matter? Plasma physics? Photonics? Binoculars? Doing pure theory is of course an option, as you can do it with pen and paper, but I rather like actual experiments.