FAQs: Starship, Starlink, and some SpaceX-conomics

Infinite money printers, Get To Mars Fast, and other fun questions

Thanks to everyone who read and offered comments on my last post. A lot of folks had interesting questions. I responded a bit in comments, but between travel and work I left a lot hanging. So, I’m collecting responses to some frequent questions here, along with further musings. I’ve broken answers into short takeaways, longer explanations, and technical bits.

In my last post, I made two points about how SpaceX’s vertical integration and dominant position in the launch market would affect the way SpaceX would price rocket slots:

1. The price of a Starship slot is going to be the marginal cost plus a markup. If SpaceX is a monopolist, the markup will be (proportional to) the inverse of the price elasticity of slot demand.

2. The marginal cost of a Starship slot is going to be exactly the net present value (NPV) of profits from a marginal Starlink satellite. This is the opportunity cost of allocating the slot to a non-Starlink payload.

I, personally, find the second point more interesting than the first. The first is just a feature of having market power. It shows up in lots of places. The second is a consequence of vertical integration across launch and orbit use. It’s always interesting when a firm is integrated across two markets, where customers in one market can be competitors in another. I can’t think of many such examples off the top of my head.

(Recent WSJ reporting reveals behavior that is, at least, consistent with profit-maximizing behavior. I don’t want to read too much into this — it’s one data point — but it’s interesting and noteworthy nonetheless.)

Anyway, FAQs.

Is Starlink an infinite money printer?

Starlinks go BRRRR! Photo credit: SpaceX.

Question: At some point, the marginal benefit from launching the next Starlink satellite should go down, maybe when the constellation is fully deployed. That would take the pressure off SpaceX, and they’ll no longer be slot-constrained. Wouldn’t that invalidate the trade-off between launching other satellites and launching Starlink satellites?

Short answer: No, the fundamental trade-off linking launch prices and satellite revenues holds even if the constellation is fully deployed.

Longer answer: Suppose SpaceX maximizes (expected net present value of) profits. What’s the optimal number of Starlink satellites to achieve that?

Whatever the optimal number of Starlink satellites is, it has to equate the cost of deploying one more satellite (“marginal cost”) with the expected present value of profits it brings (“marginal benefit”). If marginal cost is below marginal benefit, they’re leaving money on the table by not deploying that next satellite. If marginal cost is above marginal benefit, they’re throwing money away by deploying that last satellite. Even if you think the demand for Starlink’s services in inelastic, orbital congestion is real. The need to maneuver satellites to avoid collisions will eventually constrain the system’s size, and, given regulatory inaction and time, will limit launch demand too.

So. The cost of Starlink satellites puts a lower bound on the expected present value of profits from Starlink satellites. Because SpaceX is vertically integrated, the expected present value of profits from Starlink satellites puts a lower bound on the marginal cost of a launch (Starship, or Falcon, or …) launch slot. The marginal cost of a launch slot puts a lower bound on launch slot prices. As long as SpaceX is trying to increase (the expected net present value of) profits, they’re going to face incentives that couple launch prices with satellite values.

Technical bits:

1. This all goes through even if you think SpaceX is optimizing some other function of their profits, e.g. investor perception of expected net present value of the business. “Marginal benefits” would then mean “marginal utility of profits” instead, and the cost would be interpreted in utility units.¹

2. At a steady-state constellation size, replace “next Starlink satellite” with “increasing the steady-state size of the constellation by one satellite”. The marginal cost/benefit of a satellite should then be interpreted in terms of steady-state satellites maintained, not individual payload deployments. Similarly for permanent increases to constellation size while away from the steady-state size. There are interesting dynamics here, but they don’t affect this point.

What if they just want to Get To Mars Fast?

Your mission, should you choose to accept it… Photo credit: Dawn Hudson.

Question: You’re assuming SpaceX is maximizing (some function of) profits. But they’re actually trying to get to Mars as fast as possible. How does that affect your conclusions?

Short answer: Money helps with big projects. Profits help with getting money.

Longer answer: Getting to Mars and building a sustainable habitat/society there will need, to put it mildly, lots of capital (“money”). Profits are helpful in getting money, so “get to Mars faster” probably involves, at some level, “try to increase profits”.

Whatever its goals, a firm has three sources of capital: cash, debt, and equity. Cash comes from making and selling stuff. Debt comes from lenders who want to get paid back, plus interest. Equity funds come from investors in exchange for claims on future profits and/or control rights.

So if/when SpaceX needs capital, what can they do?

• To get cash: SpaceX can work to increase profits.

• To get debt: SpaceX needs to convince someone to loan them the money, which means they need to convince them they can pay them back.

• To get equity: SpaceX needs to convince investors the firm is worth investing in.

Cash is probably the most sustainable option. If they sell stuff, they can get money without strings. If the stuff they sell is useful to GTMF, so much the better. This will push them to behave like a profit maximizer, creating the launch price-satellite value connection.

Lenders like to see profits, or at least the potential for profitability. It would help in making that case if they behaved like a profit-maximizer for a spell. We’re back to the launch price-satellite value connection.

Equity is interesting. Some investors will want ROI. They’ll need to be convinced the firm will generate (probably significant) profits. Other investors will want to see SpaceX get to Mars fast. They’ll need to be convinced SpaceX can execute, which may or may not involve generating more profits (possibly for second-order reasons). To the extent they need to increase profits, we’ll see a launch price-satellite value connection.

Technical bits:

1. The GTMF optimization problem: minimize time to Mars, where time to Mars is a decreasing function of funds available. Minimizing a decreasing function of profits is like to maximizing an increasing function of profits, and the first-order condition will link launch prices and satellite values.

2. Or, you could consider the problem with funds available as a constraint. Since profits generate funds, you'll again end up with similar FOCs, just with a shadow “price” (Lagrange multiplier or costate variable, with units of time per money) scaling the importance of this connection to reflect the value of more profits.

3. But suppose time to Mars stops decreasing past a certain level of money. Then, sure, if SpaceX has enough money, optimal behavior may no longer link launch prices and satellite values. Their utility of money is probably still positive though, e.g. because money helps with the “survive on Mars” objective. Space exploration is one of those projects that can productively absorb (effectively) unlimited amounts of funding, so I expect profit-maximizing behavior will always show up.

What if SpaceX wants to grow the space industry?

Starship, tending to the garden of future space activities. Source: DALL-E 2.

Question: You’re assuming SpaceX wants to profit off of Starship, but they’ll have more profits later if they grow the long-run market for going to and using space. Wouldn’t they want to keep Starship prices low to do that?

Short answer: Maybe that pushes Starship slot pricing down for a little while, but sure as the sun rises every day, demand curves eventually slope downward. As long as they have some market power they’ll price at cost plus markup, not markdown.

Longer answer: It’s possible, I suppose, that there is a threshold of launch prices below which many new use cases are unlocked, creating a strong (revenue) incentive to keep prices just below that threshold.

The MITRE Corporation’s launch demand analysis — used by BEA — finds low-ish elasticities (below 1 in absolute value). They focus on substitution elasticities between rockets, holding total launch demand fixed, so it might not be fully transferable to what Starship would do. I don’t know. But, no threshold visible yet.

Anyway, even if there is a demand-unlocking threshold, the demand curve will still slope downwards after it jumps up. If nothing else, budgets are finite. While there could be some markdown in the short run to get under the threshold, the markup pricing rule will win out in the end.

Technical bits:

1. Pick your favorite story where the location of the demand curve is a function of past prices. A launch oligopolist’s optimization problem will give the price three components: the usual marginal cost and elasticity markup, and the value of growing the future market. As long as marginal costs are eventually increasing, the size of the market the oligopolist will want to serve is bounded, and the value of growing the future market will fall over time.

2. “The markup pricing rule will win out in the end” just means “the demand curve must eventually slope downwards, and SpaceX will eventually find themselves settling at such a downard-sloping portion of the demand curve.” Anyway, the markup pricing rule doesn’t affect the connection between launch prices and satellite values, that falls out of the vertical integration. Don’t believe me? Try solving the problem when demand has a jump discontinuity. Be sure to check the value of profits at points where the FOC is satisfied below and above the discontinuity, as well as the second-order condition.

So what’s it gonna cost, doc?

“These prices are out of this world!” Source: DALL-E 2.

Question: What do you think the price or cost of a Starship slot will be?

Answer: I don’t know. Payload Space seems to have good projections. I enjoy reading their newsletter. Check them out if you’re into this stuff. But there are a lot of unknowns about demand and competition before we can make a point prediction from the markup pricing rule.

Still, we can use the connection between Starlink satellite values and Starship slot costs — a result of vertical integration — and what we know about Starship capacity and Starlink mass to ballpark a lower bound. Payload’s numbers imply a revenue per satellite (~$890,000) close to my low case from the last post.

Starlink satellite revenue estimates with numbers above. Spreadsheet here—play with your own copy, let me know if you see a mistake.

Tim Farrar and Pierre Lionnet point out that the WSJ reporting implies a cost per satellite on the order of around $2M, about 4x as high as the $500,000 “manufacture and launch only” value I used in the last post. So the net present value of a Starlink satellite is maybe between $1-3M. Let’s take the lower bound to be optimistic about Starship.

If the marginal cost of a Starship slot ends up around $1M ($3M) because of the connection to Starlink satellite values, the price of a flight ought to be considerably higher. Say Starship can take about 150,000 kg to LEO. Say the latest Starlinks weigh about 1,200 kg. At 125 Starlinks per Starship, the opportunity cost of a fully-loaded flight is about $125M ($375M). Even with a 100% markup that’s a lot cheaper than a SLS flight, but even at cost it’s not a lot cheaper than heavy lift vehicles have historically been. It’s true that Starship is a much bigger beast than the other vehicles, but we’ll just have to wait and see whether there will be use cases that need so much capcity.

It goes on and on and on and on

Nobody knows what the future holds for Starship and Starlink. Will new space use cases materialize? What is the price elasticity of launch demand, actually?² When will Starship fly? What kinds of competition will it face? How will consumers value and substitute between the different launch vehicles available? There are similar questions for Starlink too. I find SpaceX fascinating; its vertical integration across launch and orbit use does interesting things. Like coupling profit-maximizing launch prices with satellite values.

Environmental sustainability (in orbit and on Earth) will drive long-term profitability and sustainability for SpaceX and every other space firm out there. Thinking forward, how will orbital environment evolve? How will SpaceX navigate environmental issues on Earth? Will atmospheric disposal seem like a relatively dirty disposal option soon? I think the environmental aspects of space exploration, on Earth and in space, are probably the most important and underappreciated unfolding space economics story of our time — it’s what most of my research is about. But these are subjects for another post.

Anyway, thanks for geeking out about space economics with me :)

It’s a Journey, that’s for sure

1. “Marginal equals marginal cost” is just a pithy economics-y way of saying “at an extreme point of a function, the gradient is zero” — it’s true whether the function is called “profit” or not. Don’t disagree with me, disagree with the universe or whatever deity/ies you recognize for making optimization work this way.

2. It is not the raw changes in launch price and launch volume over time. That measurement conflates shifts in supply with shifts in demand.