SpaceX, launching thinsats
This is a summary of data at http://www.spacex.com, plus some estimates for M288 and M480 thinsat launches
vehicle 
PSLV 
Falcon 1 
Falcon 1e 
Falcon 9 
F. Heavy 
Saturn5 

Pad Weight kg 
295000 
27670 
35180 
333400 
1400000 
2800000 

1st Stg Dry kg 

1360 
2580 


288000 

1st Propellant 
HTPB 
Kero/LOX 
Kero/LOX 
Kero/LOX 
Kero/LOX 
RP1/LOX 

1st Proplnt kg 
192000 
21490 
39460 


2012000 

1st ISP sec 

300 
304 
304 

263 

2nd Stg Dry kg 

544 
544 


36000 

2nd Propellant 
UDMH/N204 
Kero/LOX 
Kero/LOX 
Kero/LOX 

LH2/LOX 

2nd Proplnt kg 
40000 
4040 
4040 


444000 

2nd ISP sec 

317 
317 


421 

3rd Propellant 
HTPB 






3rd Proplnt kg 
4000 






4th Propellant 
UDMH/N204 






4th Kg 
2000 






LEO payload kg 

410 
1010 
10450 
53000 


GTO payload kg 
1050 


4680 



cost M$ 


10.9 
54 


MoreLater  Server sky estimated launch
How Cheap Can It Get?
To an outside observer, the main added value for SpaceX appears to be high automation, mass production, and relentless logistic improvement. If SpaceX can run a rocket program the way PayPal runs an online bank, they will launch a lot of payload while improving performance and reliability with a small staff. For the physical aspects of production, I hope they copy best practices from the electronics industry. A rocket is like a very large cell phone, and can be put together with larger robots. Massive amounts of flight telemetry from many flights can be assembled into reliability models, improving launch yield while reducing expensive defects.
Assume 5 grams per thinsat (including a fraction for the stack dispenser). Assume 5000kg to M288 for a Falcon 9, or 1 million 5 gram thinsats. http://www.spacelaunchreport.com/falcon9.html assumes 22 tonnes of dry mass and 280,000 kg of propellants and 6000 kg of payload. Assuming $2/kg for RP4 and $0.20/kg for LOX, and a 25%/75% weight ratio (slightly fuelrich), the propellant cost is $0.65 per kilogram of fuel. The propellant will cost about $200,000.
The energy cost of the propellant is about 15MJ/kg for the RP4 (equivalent electrical generation potential) and 3.6MJ/kg for the LOX, so the energy cost for 280,000 kg of propellants is about 2E12J. 22,000 kg of aluminum at 54MJ/kg is 1.2E12J. The total is 3.6E12J, or 1 million kilowatt hours, about 0.83 kWh per thinsat. If each thinsat produces 4 watts with 75% duty cycle, 3 watts average, they pay back their launch energy cost in ten days.
The dry weight is expensive; while SpaceX assumes return and reuse, that will require extra hardware for recovery, the weight of which comes right out of the payload fraction. Further, mechanisms wear out. Designing them for many flights will mean stronger construction, more weight, and more rework of damaged systems between flights. Metal is cheap; aluminum is $2/kg, carbon fiber is $11/kg, glass fiber is $2/kg, etc. With enough automation, the dry weight might cost an additional $300,000 , materials and labor. If SpaceX gets good at this, they might be able to sell a flight for $6M and still make an 80% profit. If they get really good at this, payloads will show up, get tested and mated and launched by robots, with just a salesperson and a range safety operator (who sweeps floors between launches). We can dream, anyway ...
That works out to $1000/kg to M288, and $5 per thinsat launch cost.
Wildly hypothetically, if we paid just marginal cost to LEO, and used reusable tethers to launch thinsat packages into higher orbits, $500K for 10,450 kg or 23,000 lbs works out to $48/kg or $22/lb . Given other costs like range safety insurance, launch pad rental, staff, etc., rocket cost will never get that low, but it is conceivable that it could drop below $100/kg at very high launch rates.
If thinsats can be made much thinner, with space debris or lunar ballast added by already deployed dispensers in orbit, the launch weight can drop below a gram and the launch cost below a dollar per thinsat.
An additional two orders of magnitude cost reduction may someday be provided by the launch loop, along with lunar glass substrates, and inorbit automated assembly and repair. In the long run, only the semiconductor die and the thin film InP solar cells need be launched from earth, with aluminum and glass from the moon. Earth launch might drop below 100mg per 3 watt thinsat.