All engineering problems are to an extent. In the final analysis, in 47 years, I've found that it's mostly moving the problem from one in-box to another in-box. The problem is still there, but now it's not your in-box.
Just like how most computers are air cooled and not water cooled.That's still fixable. Just like how most computers are air cooled and not water cooled. They could build a very large air cooling tower and not need water at all.
Taking your last point - time. Time is what you're seeing as a steady state. Will Microsoft still be interested in power in 20 years? Will they form a shell corporation to sign that contract, then bankrupt it once their needs are gone? So many business use this "one simple trick" to get subsidies or reduced rates, then almost inevitably leave the taxpayer with the sack to when it suits their profit model to exit their obligations.
Question: What do you use to power the air flow for that air cooling system, and how many watts would it draw to meet the same caloric (or BTU, if you prefer) cooling of using water versus air. It's a simple formula to work out the differing efficiencies of using which fluid, air or water, you'd prefer for non-forced draft cooling. It gets a bit more involved calculating the power needs for forced draft cooling. No matter what you do though, your cooling is going to consume a portion of your power output, reducing overall efficiency, and thus adding to the time to meet ROI. The reason engineers use PWR reactors is that it makes the business math work, and other methods to date do not.
Cooling from cheap to expensive:
Mercy snip
You left out several options:
1. Don't use WCR's. MHD or PBR are a thing too. (See final line)
2. Use Geothermal cooling. Not aware of anything outside of paper but as long as you have a sink cooler than your source, you're good. (hint: The rocks can only absorb so much heat before you've exceeded their ability to sink any more and thus overheat your installation.)
3. Dump waste heat into salt vats, rotate the salt vats to allow natural cooling. (Molten salt cooling).
4. Use micro reactors like Toshiba and Samsung have developed. Put the power plant where the power is needed.
But none of those are commercially feasible without forever government subsidies. Again, privatizing the profit, socializing the cost.