They ask investors what they are willing to pay for them. That’s kind of what syndicate desks do. Hedging such positions may be a little trickier, so you don’t want to hold a lot of such paper unless you actually want the exposure, but it also doesn’t seem crazy to assume a relatively high correlation among various very long rates.
Assume a flat yield curve beyond the last observed bonds, and price the bond off the curve.
If there were other issuers with ultra long bonds, you could build their yield curves and take the 30-100 slope, and argue why (or why not) your new issuer should have the same slope.