Thread by @LionnetPierre

What about some constellation economics for a change? I've been playing around with the data made available by @erikkulu on constellations and completed it with my own information on top. So let's have a look at the current status of commercial constellations.
1/

According to @Erikkulu there are >200 active commercial satellite constellations, in various stages of elaboration and deployment. Many only exist on paper, but quite a few (156) have secured some funding and/or have deployed prototypes, and/or initial capability.
2/

The total funding secured by the 156 constellations amounts to 17B$ approximately. Noting that the Top 10 gather 85% of the total funding, and the Top 5 are worth 70% of the total (they are: OneWeb, Iridium Next, Starlink, Globalstar 2G and SES O3b/mPOWER).
3/

These 156 projects already deployed 3800 satellites, worth 750 tons in LEO, (unsurprisingly, the Top5 represent 87% of that), and announce plans to deploy 37000 in total (and >50000 if we included all projects). Hence the current status of completion is about 10% (or 7%)
4/

If we accept that 90% of the capacity is still to be deployed, the total Capex requirement will have to grow a lot in the coming 4-6 years. At full capacity we can assume the book value of the global constellation infrastructure to be worth >>150B$.
5/

Considering a useful satellite lifetime in the order of 4-7 years (noting the disparity of situations, e.g. the 1st Iridium constellation lasted over 15 years, while the first Starlink satellites are already retired), the average Capex amounts to an average 25-30B$ annually.
6/

Considering that >95% of that mass (and 90% of Capex) are for communications services, we can start making comparisons with other telecommunications infrastructure and economic drivers.
7/

For instance, let's bring in the total Capex of US telecommunication operators, worth 90B$ in 2019. With 2019 revenues in the order of 600B$, the US telecoms sector has a revenue to capex factor of 6,6.
8/

Applying this 6.6 factor to the global constellations Capex would call for a global revenue of 150-200B$/year to support the constellations infrastructure. With an ARPU of 500 to 1000$/year we would contemplate a user base in the order of 200-300 million.
9/

Is this figure plausible? Probably not. Then, how can the promoters achieve a sufficient level of revenue to pay off their investment? The answer is to charge much more and seek higher ARPU. This means that we are still far from an affordable space communications market.
10/

Iridium and Globalstar are extremely expensive services, and they barely make it. While they post revenues in the order of 100 to 400M$/year, ARPUs are awfully low for these operators, and profitability is a constant challenge.
finance.yahoo.com/quote/IRDM/fin…
finance.yahoo.com/quote/GSAT/fin…
11/

Starlink is ramping up on customers, but, at 100k users and 120m in revenues, is still very far from significant targets, i.e. 30B$ (or 25M customers). And many have doubts that the constellation of 12000 can technically support such a revenue.
reuters.com/business/aeros…
12/

Oneweb has already gone belly up once, it has now raised more capital and is completing its deployment, but its business case seems still very fragile. And apparently the sector is wondering how this can actually work.
spacenews.com/industry-skept…
13/

How exactly can the satellite infrastructure support such a high level of expectations and attract so much capital? Can it withstand high deployment cost and constant Capex to replenish? To be attractive the constellation business case must be cheaper. Can this be achieved?
14/

If we draw a quick ratio, we note that the 750 tons already deployed represent an average Capex of 23k$/kg in orbit. But each infrastructure bears different costs. Starlink deployment is far less costly than e.g. Iridium Next.
investor.iridium.com/2019-02-06-Iri…
nextbigfuture.com/2019/12/spacex…
15/

Is 23k$/kg in orbit cheap or not? In absolute terms it is very cheap, compared to historic costs of satellite in orbit (often >>150k$/kg), or even compared to the actual costs incurred by Planet and Spire (as documented in their financial statements) that are 3x higher.
16/

But even at such a bargain price, the cost of the constellation infrastructure is unsustainable, unless it produces much higher revenues per kg in orbit or if it can last way much longer. There is no way out of this cost equation.
17/

At 20k$/kg in orbit the 37000 satellites to be launched would cost 150B$, and again 150B$ every 5 to 7 years. There is no way out of it. So how can this be sustained? What is the potential to further drive down the costs beyond that?
18/

The smallsat 'revolution' has translated into lower unit prices, at the expense of reduced performance and shorter lifetime. I am not sure that the sector has gained a definitive economic advantage, while I recognise that it has enabled a different approach to missions.
19/

Can the space sector further drive down infrastructure costs? I have serious doubts. Regarding launch services, the launch price per kg reduction we have witnessed in the last decade is already impressive (30 to 50%). I don't think it can be much better than that.
20/

Of course some will point me to SpaceX and how it is driving down costs by many orders on magnitude on both launch and satellite production. I will remain sceptical until I see the goods (i.e. financial statements) and an operational Starship, with customers.
21/

Today we find statements that Starlink satellites cost less than 300k$ each and that F9 marginal cost of operations is 15M$. This puts Starlink deployment at 2k$/kg in orbit, i.e. one order of magnitude lower than the current sector average.
nextbigfuture.com/2019/12/spacex…
22/

We also have a SpaceX COO quoting 10B$ for Starlink deployment. When this statement was made Starlink was planned with 4000 satellites, and it roughly translates into 10k$/kg in orbit (if we disregard ground segment costs).
eu.floridatoday.com/story/tech/sci…
23/

However, both Spire and Planet document average infrastructure costs of 300 to 400k$/spacecraft in orbit, i.e. 60 to 80k$/kg. These costs are not very different than the ones of the legacy space sector for commercial systems.
24/

So what is the magic ingredient of SpaceX to lower cost so much? Is a cost of 2k$/kg to orbit credible? Would a cost of 10k$kg to orbit be more realistic? I don't know, but what is sure is that at 20k$/kg in orbit the constellation infrastructure is not sustainable.
25/

If 2k$/kg in orbit becomes the industry standard, then, even with 900 t/year at launch for constellation deployment in the decade, the total revenues of the space infrastructure industry cannot be expected to grow more than 1,8B$/year (400M for launch, 1.4B for satellites).
26/

I believe that current projections for multibillion space sector growth driven by the demand for constellation deployment are not economically sustainable. Today all evidence is telling us that the sector average infrastructure cost is still too high to sustain the frenzy.
27/

So here are my key takeaways on constellations. To put it very briefly I believe that mega-constellations do not represent a major growth market opportunity for the space infrastructure market, due to the high pressure they put on costs to be economically sustainable.
28/

@threadreaderapp unroll