In 2014, OneWeb ignited a new space race which has since been joined by SpaceX, Telesat, LeoSat and recently Amazon, all seeking to blanket the planet with broadband connectivity through hundreds to thousands of low-Earth orbit (LEO) satellites. Despite the altruistic and heartfelt commitment to close the digital divide, connecting the 3.7 billion people still lacking Internet access, will not be lucrative enough to fund these multi-billion ventures. In search of revenue opportunities, the so-called mega-constellations might compete increasingly with terrestrial fiber networks.
Traditionally communication satellites have been placed in geostationary orbit (GEO) at an altitude of 35,786 km, corresponding to three times Earth’s diameter. The laws of orbital mechanics dictate this great distance so satellites can circle Earth synchronously to its rotation and so hold a fixed position in the sky for years and without propulsion. This great distance impacts performance, weight and cost of satellites. Much rocket thrust is needed to lift heavy satellites into GEO and correspondingly high transmission power is required to compensate the loss in signal strength on the long path to the ground. And, despite the fact that radio signals travel at the speed of light, it also introduces a significant round-trip delay of half a second.
Operating satellites closer to the Earth brings many benefits including lower latency, higher bandwidth and lower power requirements — which in turn allows for smaller cheaper satellites to be built and launched. However not only does the field of view towards the Earth, and hence the coverage area, shrink with lower altitudes, but the laws of orbital mechanics, in particular Kepler’s second law, dictate that satellites closer to Earth must circle with higher angular velocity in order to compensate for the stronger gravity. A satellite placed in LEO at an altitude of 1,200 km would need to travel at 26,100 kph to overcome gravity, converting into an orbital period of just 109 minutes. As observed from the ground, such a satellite would move permanently, reducing its visibility to 10 minutes in every 109 minutes orbit. As a result, an entire constellation of evenly separated satellites is required to provide uninterrupted connectivity, and user terminals need to be handed over from setting to rising satellites every couple of minutes.
Depending on these satellites’ altitudes, such LEO constellations are planned to consist of at least 78 (LeoSat’s minimum configuration) and up to 12,000 satellites (SpaceX’s Starlink in final configuration) at an estimated cost of $3 billion to $10 billion.
Put into perspective to the less than 2,000 active satellites each handcrafted over months by hundreds of engineers today, it becomes clear that the satellite industry has to undergo a paradigm shift towards automatized high-volume mass production, involving the entire supply chain, to achieve the necessary output rates. And this transformation has to happen quickly, as OneWeb, which received FCC approval in 2017, has until 2023 to launch half of its planned 600-satellite constellation and SpaceX must deploy more than 2,000 satellites by 2024 to preserve its license.
The growing population of satellites also raises fears of more collisions in orbit, leading to fields of debris that in a worst-case scenario could trigger an unstoppable cascade of collisions, known as the Kessler syndrome, which would leave LEO inaccessible for years, decades, or centuries before these objects’ orbit naturally decays enough to re-enter the Earth’s atmosphere.
Interference is also feared from and with existing GEO satellites. The launch of High and Very High Throughput Satellites (HTS/VHTS) into GEO is causing sharp declines in capacity pricing and the fact that satellite operators share the same frequency bands means interference avoidance may require megaconstellations, which have lower priority, to leave some areas where beam direction aligns with GEO (particularly the equatorial belt) with lower capacity. The fact that two thirds of the Earth’s surface is covered means that accordingly two thirds of LEO satellites will be flying over areas where apart from aircraft, vessels and oil rigs there are no users limiting the revenue opportunity.
With all these technical, economical and legal challenges, doubts have increased over the past months. The unit cost of OneWeb’s satellites is believed to have doubled to $1 million despite the fact that the company partnered with Airbus to build a dedicated factory in Florida for efficient mass production. After raising $2 billion in equity, primarily from Softbank’s Vision Fund, OneWeb officials originally hoped to fill the gap of an estimated $4 billion through debt financing, particularly from French export credit agency BPI.
Amazon, despite entering the race late, has substantial strategic advantages: the company can easily fund a multi-billion dollar project and bear a low return-on-investment as upsales from its eCommerce and AWS cloud services could cross-subsidize the satellite constellation.
LeoSat has been focusing on enterprise connectivity since inception. Each satellite will feature four laser communication terminals (LCT) to establish optical inter-satellite links (ISLs) to its neighbors, forming a meshed network in orbit, allowing connectivity from rooftop to rooftop anywhere in the world while avoiding any terrestrial infrastructure. Apart from higher security, such architecture will beat a terrestrial fiber network in latency since the laser beams from satellite to satellite will take a more direct route than meandering fiber cables, and also fiber optic is a solid body slowing down light by one third, whereas in the vacuum of space, light reaches its full speed, leading to a substantial advantage in latency.
Similarly, SpaceX has been highlighting Starlink’s ability to provide low-latency connectivity, which journalists have framed as “a license to print money” since financial institutions involved in high-frequency trading can convert any nanosecond gained over its competitors into very substantial profits. Although SpaceX’s 60 prototype satellites launched in late May 2019 do not carry LCTs – very likely due to their high cost – the ultra-low latency market seems to be a priority. Although there are several reasons why the company requested to lower Starlink’s altitude from 1,110 km to 550 km last year, one implication is that the move will lower latency over LeoSat’s constellation to orbit at 1,400 km.
OneWeb cannot compete for long-haul latency as its satellites won’t have ISLs but the company has early highlighted its ability to provide backhaul connectivity to cellular networks and support 5G rollout which could undermine the expansion of fiber networks in areas sitting on the edge of viability.
Lastly megaconstellations, in particular Amazon’s, will likely target cloud customers, in particular enterprises using private cloud connectivity, which despite Megaport and PacketFabric still remains a tedious and expensive to implement, especially outside of major data centers. Amazon’s “Project Kuiper” could actually become a global cloud on-ramp and the launch of the AWS Ground Station service last year could be an indication of AWS’ strategy to drive more business into their cloud using satellite.
Despite the many challenges, it is expected that at least two megaconstellations will materialize and OneWeb is very likely the first of them. While their focus is likely on unserved and underserved customers, there will be areas and niches where terrestrial networks will face competition from space.