The following data on Falcon 9 upper stage was retrievied by Google's Gemini AI from online SpaceX files:

• dry mass $M_d = 4,000$ kg (after jettisoning fairing early in burn);
• mass of full propellant load $M_{prop} = 111,500$ kg;
• payload mass $P = 17,500$ kg when booster is landed on downrange drone platform;
• ISP for Merlin vacuum engine is 311 seconds for $V_{exh} = 311 \times 9.8 = 3,048$ m/s.

After staging, and ignoring the 1,700 kg of the fairing jettisoned after the first few seconds, the upper stage begins firing with

$$M_0 = 4000 + 111,500 + P = 133,000\ kg$$

For $M_f$, we have

$$M_f = 4000 + P = 21,500\ kg.$$

From the rocket equation, we then have

$$\Delta V = V_{exh} \times ln(M_0 / M_f) = 3,048 \times ln(133,000 / 21,500) = 5,554\ m/s$$

Summarising, to deliver a 17,500 kg payload to LEO with booster recovery, the expendable Falcon 9 upper stage must deliver 5,554 m/s of $\Delta V$. Overall, the launch consumes 522,500 kg of RP1 and LOX propellant. That's a pqyload to propellant ratio of around thirty to one.

If it were delivering payload to an orbiting spaceport supplying 3200 m/s of $\Delta V$ to arriving flights, the Falcon 9 upper stage would only need to supply $(5,554 - 3,200) = 2,354\ m/s$. With an exhaust velocity of $3048\ m/s$, we'd have

$$M_0/M_f = exp(\Delta V / V_exh) = exp(2,354 / 3,048) = 2.165$$ $$M_f = M_0/2.165 = 133,000/2.165 = 61,440 kg$$

That's a 3.5x increase in payload and a 43,940 kg reduction in overall propellant consumption. The net payload to propellant ratio drops from 30:1 to 7.79:1, a roughly four-fold improvement.

Those calculations apply to the current expendable upper stage for Falcon 9. For a fully reusable upper stage, the improvement would less dramatic. The dry weight of a reusable upper stage (RUS) would almost certainly be more than the 4,000 kg of the current expendable upper stage (EUS). The RUS would require wings -- or at least control surfaces similar to those of the SpaceX Starship -- and a thermal protection system (TPS). The low atmospheric entry velocity of the RUS after delivering payload to the spaceport would reduce demands on the TPS, so the added weight might be modest.

The mass penalty for the RUS of wings and TPS could be offset at least somewhat by the fact that the RUS would be a new multi-use space vehicle design. Before ever delivering its first cargo to the orbiting spaceport, it would find use as an intercontinental skip-glide vehicle for USSF missions, as a middle stage for launching high $\Delta V$ deep space missions, and as a reusable sub-orbital vehicle for scientific research missions. Its propellant tanks would be less than half the size of those of the EUS, and its Merlin engine might be replaced with a lighter and higher performance derivative of the Raptor 3 vacuum engine. Overall, it seems plausible to assume that something close to a 60,000 kg payload capacity for delivery of cargo to the orbiting spaceport could be retained.