SpaceX StarLink off to rough start, loses 3 satellites, risks egg on face

SpaceX launched the first 60 StarLink satellites back in May, as part of its plan to create a global broadband network in a low-earth orbit (LEO) constellation. Not aimed directly at IoT devices, SpaceX’s plan was ambitious, and now it looks like things are not going according to plan. The company has lost contact with 3 of its new satellites, and given that it wants to launch around 12,000, that’s an unacceptably high failure rate.

Of course, the charitable take is that these are just teething problems in what is not much more than the initial test program, and that these issues will be ironed out in swift fashion. The more cynical take is that a failure rate of 5%, in a communications infrastructure world where four or five nines are considered necessary for a minimum viable product, is emblematic of the Silicon Valley ethos creeping its way into a traditionally conservative world – where ‘move fast and break things’ might get you a satellite constellation, but it could also interfere with scientific radio telescopes. Unsurprisingly, there’s been no comment from SpaceX about this.

These SpaceX satellites are very different beasts to the nanosatellites that are popular among startups. Riot will shortly publish its market forecast on the satellite IoT opportunity, but where nanosatellites weigh ten or perhaps twenty kilograms, these SpaceX units are 227kg each – despite both using the same LEO orbit. Planning to launch one or two thousand of these each year, SpaceX is catching flak from scientists that are concerned at the growing threat of clutter in earth’s orbit.

Amazon’s Project Kuiper and OneWeb both want to launch broadband services on the back of satellites, and if SpaceX is successful in launching thousands of units, then the risk of creating major debris fields will be especially concerning. Currently, about half of all space debris in orbit comes from a Chinese anti-satellite test in 2007 and an accidental collision between two satellites in 2009 – an inactive Russian unit and one of Iridium’s fleet. In that paper, the authors predict that the new constellations would generate 67,000 collision alerts each year, which would require intervention or risk a collision. Coordinating that fleet would be quite tricky.

With India’s recent, politically motivated, and somewhat bizarre, satellite killing test, after which it announced it was now truly a space power, that proportion is now likely higher, and if there’s a technical failure on one of these LEO units, a collision could be disastrous. There is currently no practical way to remove the debris, so once it is there, it is potentially going to damage further satellites in the constellation, as well as hamper rocket launches. New debris sensors, the USA’s Space Fence, will be able to track object 2cm across, down from today’s 10cm accuracy, which could reveal a much greater problem than we now anticipate. Even a bolt travelling at a few kilometers a second is enough to destroy a satellite.

SpaceX says that two satellites are going to be deorbited, so that it can see what happens when a live satellite will reach its end-of-life stage. It is hiring aggressively, so should be able to resolve the problems that have struck the 3 satellites in the initial launch. These 3 apparently made initial contact with terrestrial control, but then went dark. At their current 550km altitude, they should de-orbit naturally within five years, breaking up in the atmosphere.

The LEO orbit should provide much shorter network latency than is traditionally provided by satellite broadband. SpaceX is going to benchmark this with the first batch of satellites, but it has talked about 15ms of latency, which is around a tenth of what most providers are capable of.

With 12,000 satellites, SpaceX won’t really have an excuse for lack of coverage, but the big question remains how it is going to monetize StarLink. It might want to move against the fixed-line broadband providers, targeting both the business and consumer markets and possibly shaking up the copper-bound providers that have not yet made the jump to fiber.

In the IoT side of things, any remote application could make use of this network, so long as it is not battery constrained. SpaceX has not spoken about this, and given the crop of satellite IoT startups that want to throw down with the incumbent satellite providers that are slowly turning their attention to the IoT, SpaceX might want to stay clear of that market – or it could decide to be the king, and throw its (likely) considerable weight around.

SpaceX is uniquely positioned here, in that it operates a very competitive launch vehicle, which many of these satellite IoT startups have used to launch their own tests. None of these startups talk about launching constellations of the same order of magnitude as SpaceX, with most planning on less than 100 units. This gives you a sense of scale about where SpaceX is planning on competing – at the higher-bandwidth consumer communications, rather than low-power, battery-constrained IoT devices.

This week also saw Alphabet’s Loon announce that it had broken its record for longest concurrent flight, for its balloons that power its own communications network. Balloon P-496 stayed aloft for 223 days, having circled the globe once and then spent most of its time in position, west of South America, testing Loon’s ability to keep the balloons in relatively geostatic positions.

Loon has deployed its network to support humanitarian disasters, in Peru and Puerto Rico, but it is now entering its first commercial trial, in partnership with Telkom Kenya. Loon will be serving a collection of mountainous villages, pending final approval from Kenya’s aviation regulator, which is expected within the month.