I had estimated that by about the 2500s, we'll have to throttle back world fusion production otherwise climate change will occur by sheer force of output power being a sizable fraction of solar output.
They'll then demand a gazillion in subsidies to put up giant fans to cool the planet. (Wait a mo, they could just hook up those ancient wind turbines that are still littering the hillsides as motors..)
Seriously, a common misconception is that fusion is incredibly hard, and only happens in massive machines. Fusion is ridiculously easy to do. It can be achieved with table top apparatus. Look up the Farnsworth fusor as an example.
The problem isn't achieving fusion, it's getting more energy out than you put in. Part of the problem is that hydrogen has a very low density, and that means the frequency of collisions in a plasma is relatively low. Add to that, the positive charge means the protons repel each other strongly, and will likely just ricochet off if the collision isn't exactly inline. Thus a lot of energy is wasted firing atoms at each other to no good effect.
Compare that with fission using neutrons, which have no charge and so are not repelled from a uranium nucleus. You can see why fusion is harder!
To get a decent return -as in sufficient productive collisions- you need to compress the fuel before heating it. The sun overcomes this by way of compressing its fuel with massive gravity, a weapon does so by using the energy from plutonium fission. Neither of these is very practical for a powerplant.
An interesting idea is the Polywell, which uses a pulsed magnetic field to compress a plasma, in an arrangement not unlike the Farnsworth electrostatic fusor. One advantage is that this is a relatively small and inexpensive machine (say a 1m sphere) so definitely worth a try.
Then again there are the 'cold fusion' experiments which tantalisingly suggest that there may be easier ways to achieve the compression, by constraining hydrogen atoms inside the crystal lattice of metals.
In terms of fuels, tritium (along with deuterium) is the easiest to fuse, but has the disadvantages of being rare and costly, and of giving off very energetic neutron radiation which would likely damage the apparatus. Deuterium alone is somewhat harder to fuse, but inexpensive and plentiful. It also gives off neutrons, although not of such high energy.
The near-ideal reaction would be between hydrogen and boron, which gives off little or no hard radiation, but needs a lot of energy to make it work. Far more than current machines can apply.
The cost aspect also bears comparison with present efforts at wind and solar deployment. ITER will probably cost somewhere around $30 billion. The current annual wind and solar spend is between $300 billion and $500 billion a year, depending who you ask. So, it wouldn't need all that much scaling-back of renewables expenditure to pay for ITER right now, instead of having to wait a good few years to have the full funding.
Presently, wind and solar supply around 4% of global electricity. When you extrapolate that $500billion pa to how much it would cost to go 100% wind and solar, you see that fusion research is actually a very, very cheap option by comparison.
In view of the potential benefits of success, I think it's crazy not to allocate adequate funds to fusion research.