Wednesday, June 30, 2021

Elon Musk Starlink Update MWC2021

Elon Musk joined the Mobile Wireless Congress for a thirty-minute keynote address. His online presentation did not reveal much new information about Starlink but did provide confirmation and consistency from other remarks and with regulatory filings available to the public. 

The full video is available

Statements in quotes below are paraphrased from what I heard Elon say, and are not literal quotations. Comments following each statement reflect my best understanding, which could be seriously in error by my mistake or misunderstanding. Thanks for any corrections. 

“With over 1500 satellites in orbit, the solar power generating capacity is over 5 MW.”

“1500 satellites can deliver about 30 Tbps of data.”

Which equates to 20 Gbps per satellite, which matches earlier statements. For comparison, one Viasat 3 satellite delivers about 1 Tbps.

“Global coverage in August, except the Poles”

A fair statement in that there is satellite coverage globally but fails to account for the coverage gaps due to excessive distance from a gateway (without intersatellite links, ISL), for licensing and for available in-country facilities.  

While these issues can be resolved, they are not currently.

“Spot size is large compare to 5G, more suitable for sparsely populated, low-mid density”

Spot size is something between 14-22 km across. 

“Expecting as many as 500,000 subscribers by June 2022”

Setting aside customer acquisition and subsidy, Starlink expects to receive about $100/month/subscriber, or about $1200/year/subscriber.

500,000 subscribers * $1200 = $600,000,000 annual revenue June 2022.

“Cash flow positive will take $5B to $10B. Investment will continue, with $20B to $30B lifetime costs foreseeable”

Just for sense of scale, assume subscribers double every year with annual subscriber revenue shown. This is a really crude analysis just to see what might be possible. 

Year 1:   500,000  subscribers $0.6B

Year 2: 1,000,000 subscribers $1.2B

Year 3: 2,000,000 subscribers $2.4B

Year 4: 4,000,000 subscribers $4.8B

Year 5: 8,000,000 subscribers $9.6B

Or where subscribers up by 50% per year 

Year 1:   500,000  subscribers $0.6B

Year 2:   750,000  subscribers $0.9B

Year 3: 1,100,000 subscribers $1.3B

Year 4: 1,600,000 subscribers $2.0B

Year 5: 2,400,000 subscribers $3.6B


Or where subscribers do not emerge as anticipated

Year 1:   200,000  subscribers $0.3B

Year 2:   400,000  subscribers $0.5B

Year 3:   600,000  subscribers $0.7B

Year 4: 1,000,000 subscribers $1.2B

Year 5: 1,500,000 subscribers $1.8B


For comparison, Viasat serves around 600,000 subscribers plus about 1500 airplanes, with about $0.87B annual revenue. 

Hughes serves about 1,500,000 subscribers with annual revenue of about $1.8B

“Service in 12 countries. Cellular backhaul announcement coming. Two national carriers. Others in-work. Some 5G licenses require service to rural subscribers, represents 3-5% of the subscriptions.”

 What has not been apparent is to what degree Starlink can provide differentiated service levels. Certainly attainable, the question is how the network will respond as contention ratios elevate.

“User Terminal, the Satellite, and the Gateways; all are evolving. Starlink chose to embrace emerging technology instead of falling back on legacy equipment with less capability but higher confidence in performance and reliability, especially in orbit.” 

Aim high, work hard, accept setbacks, don’t worry about the collateral damage.

By all accounts, Starlink is an exceptional technical accomplishment. 

The emphasis has been to get a working constellation in place, mass produce an initial user terminal at a loss, and build the ground network progressively as licensing and customers direct.  There is no revenue without a service, there is no service without all the pieces in-place, even if each piece is not perfect itself.

“Latency is similar to fiber and 5G. Goal is less than 22 msec. No perceived lag, supports gaming.”

“Most sophisticated phased array technology on the satellite. User terminal uses phased array.  Handoff in microseconds, not perceptible, doesn’t budge latency or jitter. One satellite has many different spots on the ground. Digital phased array.”

Digital beamforming provides for any number of simultaneous beams without significant loss in aperture gain applied to each beam using a single set of elements. The tradeoff is in processing power and bandwidth.

For example:

~20 Gbps per satellite

250 MHz channel ~400-600 Mbps per channel

7 freq. colors (1.75 GHz bandwidth)

~ 40 channels (20 Gbps/satellite / 500 Mbps/channel)


Which equates to 40 or less beams.

It would take about 8,000 spots to service the entire satellite coverage area, which would deliver less than 3 Mbps to each spot. The beams would hop from location to location, dwelling long enough to service the clients. As many as seven beams could stack up on any location for about 4 Gbps peak capacity, leaving maybe 30 beams to cover the remaining territory.

“The user terminal recurring cost apparently is hovering at $1000, with a plan to get that to around $300.”

Digital beam forming can be accomplished with great economy once the foundry is cast and it is simply a matter of assembling circuit boards.

Constellation density can drive the scan range to less and less, which allows the least “scan loss” with a fixed array. It also permits simplifications that would not stand up if the array were not restricted in scan range. 

“The user terminal can be set up in five minutes. Point it at the sky and plug it in, in either order. Customers will use the terminal in remote locations that cannot be serviced by a technician, it must be accessible to the user.” 

The complaint I hear most is that there is no connector to the antenna – the cord is cast. This just creates issues when passing through a conduit or otherwise managing the transition from outside to inside. The antenna has a clever mechanical erecting drive that apparently is used for setup, not tracking. The existing constellation congregates in the northern latitudes, hence the emphasis for subscription in those areas as the constellation built out. It can be difficult to clear enough sky in many locations.

“Coming this year is Satellite 1.5 with laser Intersatellite Links (ISL). These links will provide for coverage in the polar regions (Arctic and Antarctic).  Satellite 2.0 will come next year with more capability. All satellites will have ISL going forward.”

Iridium emerged with RF Intersatellite Links, and it offered amazing features that saved it from the scrap heap (government station in Hawaii that could connect anywhere).  Laser ISL is much more challenging and capable. The first ten satellites launched in polar inclined orbits have ISL (since Jan 2021). Public statements have been encouraging, no setbacks noted.

“Gateways and POPs evolving into major server centers. Edge cache. minimize latency and jitter.”

With the satellites and user terminal developments in-hand, scaling the service depends on the ground segment. Starlink approach is sensible and understandable, the question is execution and prowess.

“Starlink is a complement to fiber and 5G”

This was repeated over and over, a talking point that Starlink was a friend, not a competitor.

There is no question that satellite communications is a valuable adjunct to fill gaps in coverage and for rapid deployment. The issue has been cost and performance. By pricing the home subscription at $100 world-wide (plus taxes), it can be argued that Starlink is professing service costs in the $0.35/GB range. 


For comparison 

Satellite DTH ~$3/GB (2-5 GB/month/user)

Mobile phones ~$0.40/GB (3 GB/month/user)

Cable Internet ~$0.07/GB (300 GB/month/user)

Latency is apparently in hand. 

The remaining issue is capacity, which gets to the mandate to serve low to mid density regions.

“GEO can’t serve that much bandwidth and has unavoidable high latency”

Starlink has eclipsed GEO capacity seemingly in the blink of an eye, and it has hardly gotten going. Latency is a matter of physics. But in the end, cost is king. GEO can compete on cost if they so choose. The software defined satellite and the mega-satellite are both approaches to the problem. Differentiated services and user terminal flexibility will also allow GEO some room for exception.


·      2/3 of all payload to orbit 2020

·      80% of all payload 2021

o   12% China

o   8% everyone else

·      Reusable booster and fairing

·      Lowest cost mass to orbit

·      Spacex kryptonite hall effect thruster


·      SpaceX is 20 years old. Make life multiplanetary.

o   Go to the Moon, then to Mars.


·      60% of the cost in first stage

·      10% in the fairing

·      20% upper stage

·      10% recover, launch, refurb


·      Starship

o   Biggest rocket ever

o   2x Saturn V thrust

o   Full and rapid reusability

o   100T to orbit

§  Possible to 150T to orbit

o   Zero refurbishment between flights, like an airplane

o   Lower cost propellant

§  3.5:1 mass oxygen to fuel

§  Methane, lowest cost fuel

o   Pressurization: oxygen and methane gas instead of helium (falcon 9)

o   Launch Marginal cost of $2mm

o   Add in orbital refilling – one transfer propellant – primarily oxygen – go on to Mars with 100-200T.

o   Base on moon and city on Mars

o   First orbital launch in next few months


·      SpaceX is expanding beyond Earth


·      life is good on Earth, sustainable

o   Tesla

o   Starlink

o   Neuralink


·      Expand the scope and scale of consciousness.

·      Better to be able to know what questions to ask...

...about the answer to which is the Universe



Stay tuned!
Peter Lemme
peter @
Follow me on twitter: @Satcom_Guru
Copyright 2021 All Rights Reserved
Peter Lemme has been a leader in avionics engineering for 39 years. He offers independent consulting services largely focused on avionics and L, Ku, and Ka band satellite communications to aircraft. Peter chaired the SAE-ITC AEEC Ku/Ka-band satcom subcommittee for more than ten years, developing ARINC 791 and 792 characteristics, and continues as a member. He contributes to the Network Infrastructure and Interfaces (NIS) subcommittee developing ARINC 848, standard for Media Independent Secure Offboard Network and ARINC 688, standard for Cabin LAN.
Peter was Boeing avionics supervisor for 767 and 747-400 data link recording, data link reporting, and satellite communications. He was an FAA designated engineering representative (DER) for ACARS, satellite communications, DFDAU, DFDR, ACMS and printers. Peter was lead engineer for Thrust Management System (757, 767, 747-400), also supervisor for satellite communications for 777, and was manager of terminal-area projects (GLS, MLS, enhanced vision).
An instrument-rated private pilot, single engine land and sea, Peter has enjoyed perspectives from both operating and designing airplanes. Hundreds of hours of flight test analysis and thousands of hours in simulators have given him an appreciation for the many aspects that drive aviation; whether tandem complexity, policy, human, or technical; and the difficulties and challenges to achieving success. 

No comments:

Post a Comment