The LHC tube is 27km long but ~6cm in diameter. It is true that they are pumping down a more impressive vacuum volume though, but not in the tube, but around each of the many, many superconducting magnets. That's about 9000m³ (compared to the ~150m³ of the actual tube).
Still, the LHC is hardly maintaining that vacuum continuously all year round. And it is one of the most sophisticated engineering and scientific projects attempted in history. Hardly a good idea for a train project.
It was the most sophisticated engineering not because of the long vacuum, but because of the tolerances, the semi-novelty of the superconducting coils at the time, the sensitivity of the equipment needed, and the extreme data collection capabilities needed to capture and store as much information as possible on picosecond timescales. None of these factors are necessarily applicable to a hyperloop.
Even so, once you've built one LHC, subsequent ones would be substantially cheaper because the primary difficulty is building the scientific and engineering knowledge needed while you're building it. Similar factors apply to programming: the first version of your program took awhile to build, but if you suddenly lost all of the code, rebuilding it would take a fraction of the time because you don't have to build your understanding of the domain the second time around.
Maybe all of these factors combined still wouldn't be enough to make hyperloop economical, but that conclusion is not obvious.
Still, the LHC is hardly maintaining that vacuum continuously all year round. And it is one of the most sophisticated engineering and scientific projects attempted in history. Hardly a good idea for a train project.