D is already cheap and even used by fission reactors. Extracting T from the blanket is presumably inexpensive, and one of the things ITER will test, but that’s an unknown. https://en.wikipedia.org/wiki/Breeding_blanket The real question is how expensive the physical reactor is going to be to build and maintain and does that offset the other savings. It’s expected for that cost to drop over time which is why it’s possible to eventually be cheaper than fission.
Finally, DT is easiest to achieve at a 50/50 ratio but DD fusion still takes place. So a lower mix of T is viable once very high Q values are possible, thus eventually zero T designs should be viable.
If you need a bunch of people with PhDs to operate a reactor, then you won't have a reactor. Not only does it mean that staffing is expensive and difficult, but also that it won't be reliable or predictable. If it is predictable, then steady state operation should be offloaded to computers.
This surprises me, for a number of reasons.
A fission reactor can be operated safely by a bunch of people with baccalaureates, whereas a fusion reactor will need PhDs, as I understand things. Also, its capacity factor will not be as good as that of fission reactors.
The waste from fission reactors can be made pretty small by reprocessing and in-reactor transformations. The radioactive waste from fusion is whole reactor vessels, which are large and difficult to handle (=expensive).
At commercial scale fusion reactors need ancillary fission reactors to manufacture the tritium they require. So you have two reactors instead of one.
Other factors look similar, except that fusion will carry an investment risk premium because of its novelty and complexity.
So under equivalent regulatory regimes it seems to me that fusion would cost as much or more than fission.
Can you explain why it might eventually cost less?