Will SpaceX become the world’s biggest telecoms provider? Probably.

By launching 11,943 satellites SpaceX will do to telecoms what WhatsApp/Facebook Messenger did to SMS and in doing so capture a $1tn+ business — and there’s fringe benefits for Tesla.

(This is a followup post to three earlier posts on forecasting. The first in May 2015 forecast both blimp-based and dedicated building-based drone deployments (later patented by Amazon); The second in October 2015 largely predicted Elon Musk’s Tesla Masterplan Part Deux by 9 months, the third in July 2016 among other things correctly hypothesised the use of Model X falcon wings for future possible Tesla bus designs. I try to get it right but I mainly enjoy the idle speculation).

NOTE: I have followed up this blog post with a yet more detailed product analysis here.

SpaceX launching Intelsat 35a


I was recently in San Francisco and had a very random number of drinks with two very friendly employees of US telco AT&T. As is often the case I turned the conversation towards autonomous vehicles, and more specifically two of Elon Musk’s companies, Tesla and SpaceX.

I was curious about how cars, such as a Model S, have much greater data connectivity needs than ever before. Right now, Teslas connect to AT&T’s network and it seems clear that data needs will only increase for data hungry vehicles that drive themselves. Already Tesla cars consume quite a few gigabytes of data per month.

Not only do vehicles like Teslas need high-end GPUs for some of their basic self-driving features, but they need good data connections for sharing data with Tesla, and for things like in-car infotainment systems.

As we move towards autonomy (specifically electric vehicle autonomy), cars will likely have even more data needs. Will passengers in autonomous Tesla Mobility vehicles expect Wi-Fi with fast broadband speeds? Will there be two, three or five people per vehicle? Remember all that time you had at least one human concentrating on driving? Well that person will instead be browsing the web while listening to a streaming music service.

This made me wonder: Why would Tesla would continue to bother with a middle-man like AT&T in the future?

Musk is a big fan of full-stack approaches to enterprise — so it seems logical to speculate that the growing data needs of autonomous Teslas will become a core issue Tesla will want to own or control (and perhaps charge for).

This brings me to SpaceX.

Musk often says he has a first principles approach to thinking. So let’s do it.

If you wanted to connect everyone on a planet to high-speed broadband, given technology available today (and not technology based on copper-based infrastructure that has been with us for decades), would you bother digging holes, laying copper, fibre, backhaul, exchange buildings, lay cable that crosses oceans, then on top of that build millions of cell towers with relatively limited range that span continents in order to connect mobile devices?

Probably not — particularly if there was an alternative available where you could grab internet from the sky.

Or as Patricia Cooper from SpaceX said:

“…the common challenges associated with siting, digging trenches, laying fiber, and dealing with property rights are materially alleviated through a space-based broadband network”

(And if you were the sort of person who wanted to build high-speed connectivity on a brand new planet — such as Mars — because the 1 million person colony there will probably want to have iPhones, you would certainly not have the approach of building a land-based communications network. You would do it in orbit. It’s much cheaper.)

As of the end of 2016 there were about 1,459 satellites in orbit around the Earth. SpaceX last month announced a plan to launch some 11,943 satellites — multiplying orbital vehicles by an order of magnitude.

Why so many satellites? Global broadband. Why is this interesting? Because it is completely independent of the switch-based POTS legacy we still largely work with (albeit we have been transforming it from a switch system to an IP-based network for decades now).

What will this new SpaceX satellite system look like? First is the LEO or Low Earth Orbit constellation of 4,425 satellites. Here’s the breakdown:

The second part of the SpaceX constellation will be the VLEO, or Very Low Earth Orbit network, comprising 7,518 satellites. By operating closer to the ground, SpaceX say this will both boost capacity and reduce latency in heavily populated areas:

SpaceX say this should give 1Gb broadband with relatively low latency (25–35ms because of the lower orbit than other space-based systems) to millions (or billions) of subscribers using relatively small user terminals with coverage across continents. They illustrate the relative coverage of the LEO and VLEO systems here:

Both these networks would be meshed to co-ordinate and provide redundancy and capacity, for constant near-global coverage. As you can see here, the higher dots are LEO and the lower ones are VLEO.

But I love the phrase SpaceX use in the document, because it’s part of the narrative companies like them and OneWeb are using about providing broadband to poorer countries or under-served rural locations:

SpaceX has designed its V-band system to meet the dual requirements of the world’s broadband demand — namely, connectivity for rural, remote and hard-to-reach end-users, as well as efficient, high-capacity connectivity at all locations.

In journalism this is called burying the lead.

What SpaceX are actually seeking is to replace every broadband and communications provider on the planet, by cutting out the middle man of land-based networks that stand between you and the internet. In doing so they will be essentially competing with every communications provider in the world — a business valued at over a trillion dollars. Forget about poor communities in Africa for a second: this is a pitch to replace physical fibre/cable connections in modern industrialised economies.

But a few questions arise from this including the big one: mobile phones. Will the plan be to have mobile phones work directly with satellites overhead? Is that even possible? Or will there be a hybrid approach — provide broadband to physically static locations and work from there?

Which brings me back to Tesla.

Musk was keen to announce recently the launch of the solar roof product — solar cells embedded in most of the tiles on your roof. This roof would in turn be connected to a Tesla battery, storing energy during the day for use at night (and perhaps for your electric car too) — and leading to relative energy independence for many users — and a distributed grid.

But this raises another interesting hypothesis in my mind. OneWeb, a rival to SpaceX for this race for global internet, have shown their prototype device conceivably connecting schools and medical centres in Africa:

What’s interesting about this? The terminal on the roof is:

  1. Solar powered — so independent of the energy grid
  2. Doesn’t just provide Wi-Fi — it has 3G and LTE built in too.

So what would I do if I were in Elon Musk?

  1. I’d sell SpaceX user terminals to Tesla Energy customers. Each solar roof (or roof with Tesla solar panels) would come with an optional extra — solar powered high-speed broadband (in a form perhaps similar to OneWeb, or even under your roof). Your internet would become independent of the energy/telecoms grid, inexpensive (SpaceX reckon $200) — and de-centralised.
  2. I’d make each device at a user’s home a base station for providing Wi-Fi, 3G and 4G coverage. I’d make the offering so cheap that demand would be enormous. (Each base station may not have a huge range given its size and power consumption but one could imagine a mesh system working in a co-ordinated fashion). If you can get say, 50m homes between the US and Europe and charge $50 a month, that’s a clear $2.5bn in revenue a month — while utilising a relatively low maintenance satellite network.
  3. With enough customers in high density population centres my own customers would be paying for the roll out of SpaceX global mobile broadband infrastructure — I’d cut out the middle man of current telecoms incumbents that rely on land-based cables and towers. (I might also build some towers too, but this would be much cheaper because they would be also solar powered and would not need a physical connection). (note: another fringe benefit of the Boring Company might also be laying fibre). In addition I’d be selling backhaul connectivity since lasers in space can send data faster than through cables.
  4. I’d then also have a global infrastructure for Tesla vehicles to be permanently connected to space-based broadband (depending on how phased arrays evolve this could be that each vehicle is connected constantly to a satellite, or it uses a mixture of that plus terrestrial devices mentioned in 2).
  5. Use the enormous revenue generated from displacing incumbent telecoms players (A back of the envelope calculation says the major telcos combined value is over $1 trillion+) and take what I’ve learned and apply the same technology to investing in, and rolling out, Mars-based orbital telecoms infrastructure.

SpaceX plan to launch their first prototype later this year and the first satellites for the constellation in 2019 with a target to complete the network by as early as the mid 2020s (but likely later). And with their success in re-usable first-stage rockets, they have the means and will to meet an ambitious satellite launch schedule, and at a cost that is likely to continue to fall.

It also might bring about the death of land-based networks such as AT&T (current market cap $223bn).

And perhaps even more ironic — SpaceX is charging existing telecoms and TV providers such as Iridium lots of money to launch satellites. But SpaceX’s longterm strategy could be to use the profits from these launches to replace their own customers with a more compelling, fast and global space-based mesh network and offer connectivity directly to every citizen on the planet. So like solar roof customers paying for SpaceX infrastructure, so too would existing communication companies pay for their own possible demise.

So in summary—

  1. Start with first principles
  2. Go fully vertical
  3. Choose distributed over centralised
  4. Let your customers pay for your infrastructure roll-out

(Disclosure: I’m a *very* small shareholder in Tesla. I’m the founder over at Vizlegal (in Ireland!) where we’re building a global API for law — a sorely needed thing if you want autonomous machines (and even a Musk Mars colony needs laws too!). I’m on Twitter if you have any questions)