That's 7 satellites per launch at the current SpaceX cadence.
Another poster said their test satellites were 400kg, and the falcon 9 can hoist 22,800kg to LEO according to Wikipedia. That gives a ratio of 57 satellites per Falcon 9. The final satellite design might be heavier or lighter, the desired orbit may require more fuel and less payload, the satellites may require some additional hardware for securely storing and deploying them, and the current F9 may be superseded by bigger or more efficient or cheaper per payload kg rockets over the course of the next few years, so 70 per launch is probably a good guess to within an order of magnitude.
If they wait for the BFR, that (also according to Wikipedia) has a projected payload to LEO of 150,000kg or 375 test-satellite-equivalent masses. If it's cheaper per kg, they might just use that.
People are currently think they can send more then 10-20 on one flight.
To minimize costs, I think it makes more sense for them to do RTLS than barge landings, especially as barge landings put hard limits on launch frequency (the barges aren't very fast).
However, even that 13,000kg is probably too high, because that is to an orbit that is probably lower than the 1150-1300km ones they are going to use for this.
However, for practical reasons enlarging the fairings can be more expensive than just flying more, so the limit of satellites per launch is probably set by how many they can fit into the standard F9 payload fairing. (1).
> If they wait for the BFR
They can't wait for very long, or they might lose their spectrum allocation. The contract requires them to put up half of their constellation in 6 years.
(1) page 36 of: http://www.spacex.com/sites/spacex/files/falcon_9_users_guid...
Falcon 9 payload to LOE is 22,800kg.
Of course, the high payload of the boosters and the low weight of a single satellite in this constellation will mean that SpaceX will launch a lot of them per one rocket.
Starting with the launch of Bangabandhu-1 on top of the B1046 core on 24th of this month, SpaceX will start launching satellites on top of the new Block 5 versions of their boosters. They are not designed to be reused once or twice with teardowns that approximate what was done with the Shuttles between the launches, they are designed to be reused "many" times, with refits every 10th launch.
What blows my mind about SpaceX is not just that costs are going to come down, it's just how much the worldwide space launch capacity is going to go up when they continue mass-producing rockets without discarding them, growing their fleet of launchers by ~1 every two months. (Factoring in the customers who want the performance of flying expendable).
Now the only thing I can think of is their new nukes.