Maybe you would like to combine two sources if you really want to do this with waveguide theywould be rather large for this frequency)you can use a "magic tee"
To combine two sources as Fritz Caspers describes, they would need to be phase locked to each other, because if they weren't, then the power would wander from one output of the T to the other and back again as the phase between the two sources varied. You would also need to get the phase right, probably with a phase tuning section, so that all the power came out of one port of the T, rather than a bit from each, or all from the wrong one. You could use any 3dB hybrid coupler, but a magic T in waveguide as Fritz Caspers suggests is probably the most compact for high power. You will need to load the other port, and with a high power load because, for instance, it will get half the power of one source if the other one fails, or all the power of both if the phase locking goes wrong!
I could also add that if the sources are magnetrons, it isn't a good idea to drive them in parallel from a single power supply, because it is very hard to get them both to start. The first one to start may limit the voltage rise and stop the other from starting, unless they are very closely matched.
It is easy to combine two independent microwave sources of similar frequency in a cylindrical or rectangular chamber using polarization decoupling. Even three such sources can be combined in a rectangular chamber. At the same time, they will be really independent - neither in phase nor in time.
Without changing the design of the microwave installation as a whole, you can simply increase the power of the generator. The two-phase (anti-phase) scheme of the high-voltage power supply makes it possible to almost double the microwave power of the magnetron. But the ripple factor remains almost 100%. A three-phase power supply scheme will increase the microwave power by 2.5 - 3 times with a ripple of 15 - 20%. In both cases, additional magnetron cooling is required, preferably liquid.