You have to look at all of the costs associated with a program, not just the marginal launch costs, when determining what it'll cost to put something in orbit. Consider a hypothetical booster capable of putting 20,000 KG in LEO. If you spend $20 billion developing a vehicle with a projected operational life of 1000 mission across the fleet (wildly optimistic IMO, given the Shuttle experience), then your R&D costs alone are:

$20,000,000,000 R&D divided by
1000 missions equals
$20,000,000 of R&D costs amortized per mission divided by
20,000 KG equals
$1,000 per KG.

That's not even counting the marginal operations costs (fuel, training, mission prep, flight ops, recovery, etc). No honest accounting for launch costs can be made if you don't factor in the R&D costs.

Take a look at what is perhaps the most economical booster right now, the good old Soyuz (SL-04). It isn't a simple design, but most of the R&D was amortized long ago. Counting the Molynia variant (adds an upper state), somewhere around 2,000 of them have been built and flown over the decades with a reliability factor over 95%.

We need to look at the bigger picture. There may be a need for 3 classes of vehicles to support large scale space exploration.

1. We need a vehicle to carry humans between the surface and LEO. It might be designed to carry a crew of say 4-7 people along with perhaps a few thousand KG of supplies. It may or may not need to glide to a landing. Because lives are at stake, very high reliability is paramount. Reusability of the crewed component would be a good idea, so long as the R&D costs aren't so excessive as to drive up the mission costs to exorbitant (no pun intended) levels.

2. Unmanned booster for low value payloads (e.g. propellant, consumables): This might use an approach that some advocate called "Big, Dumb Boosters" (BDB). A BDB is designed for easy and inexpensive manufacture, low development costs, and an acceptable reliablity rate. Since the payloads aren't very valuable, you might be willing to live with a 75% reliability figure if it results in a sharp decrease in R&D costs.

3. Unmanned booster for high value payloads: This vehicle may not need to be reusable. It might be an upgraded BDB with things like redundancies built in to increase the reliability rate.