A SpaceX Falcon 9 rocket launches a batch of Starlink satellites Dec. 18 from Vandenberg Space Force Base, California. Credit: SpaceX via Flickr

In the late 1990s, I spent three years advising Teledesic on the business plan and customer requirements for the first LEO broadband satellite constellation. We had hoped to serve millions of small businesses and high-end consumers with a cost-effective broadband solution for suburban, rural, and remote areas. However, the Teledesic project was canceled during the dot-com bust when Craig McCaw could not convince himself that the proposed $10 billion satellite system would deliver on its business plan. The terminals were too expensive, and it was far from clear that traditional satellite contractors like Boeing and Motorola could meet either the timescales or budget.

Other LEO projects, from Iridium, Globalstar and Orbcomm, were completed but quickly ended up in bankruptcy because the market for satellite phones and what we now call Internet of Things connectivity failed to grow fast enough to pay for the enormous cost of those constellations. However, these systems ultimately were saved after writing off almost all their development expenses.

Twenty years later, a host of new LEO broadband systems are coming to market, led by Starlink, and the question must be asked: will this time be different?

In one respect, this time is different. Starlink has entered commercial service, albeit with a partially completed constellation. In contrast, Teledesic never got to discover whether potential customers would line up to buy LEO satellite broadband connectivity. To date, Starlink has attracted considerable interest, with 140,000 subscribers using the system and over 750,000 people putting down a $99 deposit as of November 2021. However, Elon Musk has also told the SpaceX team that the business plan for Starlink version 1 is financially weak, and a much larger version 2 system (with as many as 30,000 satellites) will be needed to produce a strong business plan with sufficient capacity to accommodate millions of customers. Because SpaceX’s current Falcon rockets can’t launch sufficient mass or volume to accommodate what may be much larger, higher capacity version 2 satellites, SpaceX now also needs to bring its Starship rockets into operation as soon as possible, introducing additional risk to completing the Starlink version 2 deployment.

We shouldn’t discount the SpaceX team’s likelihood of overcoming these technical challenges. They have repeatedly shown the ability to improvise to overcome obstacles and build a minimum viable product that can subsequently be improved upon. Starlink version 1 is a technological milestone, and just like Iridium’s satellite phone network in 1998, it is amazing that it works at all. And SpaceX has mitigated some of the cost challenges that Teledesic faced, namely by vertically integrating to build launchers, satellites, and terminals itself. But Starlink does face specific technical limitations, such as tree obstructions due to its use of a low minimum elevation angle (recently lowered from 40 degrees to 25 degrees) and regulatory constraints preventing re-use of certain Ku-band user link frequencies (which therefore restrict system capacity in a given area).

More importantly, Starlink must also meet and keep pace with terrestrial broadband expectations in order to create a viable business — something competing satellite broadband systems have failed to do for the last 20 years. Unlike the innovations represented by SpaceX’s rockets and Tesla’s electric cars, Starlink is not entering a relatively static market where competitors seek to maximize profits from their existing products. Instead, Starlink is competing in a broadband market where data consumption per subscriber has grown by at least 20% to 30% each year for the last two decades and shows no sign of slowing. Although Starlink offers much better speeds than existing geostationary satellite broadband systems today, it is far from clear that the system will be able to provide 5 to 10 times more capacity per subscriber at the end of the decade. Governments are also pushing for ever-wider deployment of terrestrial fiber and 5G technologies. This push will restrict demand for satellite broadband to increasingly remote areas. Even existing consumer satellite broadband players, such as Viasat, now believe that substantial further growth in this segment is unlikely. As such, they are moving to diversify into government and mobility markets, as demonstrated by Viasat’s recent acquisition of Inmarsat. But these incremental opportunities are relatively small, and success in the consumer market will be critical if SpaceX is to support its current $100 billion valuation (which largely rests on the promise of Starlink) and justify investing tens of billions of dollars in the Starlink version 2 system. Arguably, Starlink’s most advanced competitor, OneWeb, is only in a position to move forward to full commercial deployment because it has already written off over $3 billion of initial development expenses.

SpaceX is fortunate in that it has had easy access to capital, raising a total of $5 billion from investors since the beginning of 2019. To date, this has provided ample funding to develop both Starlink and Starship. But a similar scale of fundraising will need to continue for years to come because there is no clear commercial demand for Starship launches, and SpaceX (even in Musk’s view) will need to launch many thousands of additional satellites — and “consume massive capital” to ramp up terminal production — before it has a chance of creating a strong business plan for Starlink. If the financial markets don’t cooperate, it may become impossible to keep all these balls in the air and raise money for SpaceX at ever-increasing valuations. Perhaps Musk himself might be able to provide some funding. However, after taxes, his recent sales of Tesla shares have only netted him a few billion dollars. Given his focus on going to Mars, it seems all but certain that in the event of a funding crunch, Starship would take priority over a Starlink system that was still struggling to ramp up and still needed significant additional capital to expand.

As a result, all roads lead back to whether Starlink really can attract sufficient customers to become a sustainable business and support the launch of the five or six thousand satellites each year needed to maintain a version 2 constellation of 30,000 operational satellites. Although securing 750,000 customer deposits may signal that many people are disappointed with their current broadband choices, Starlink’s growth to only 140,000 subscribers by November 2021 is rather less impressive. These numbers are not much better than the progress achieved by WildBlue after its (U.S.-only) satellite broadband service was launched in 2005. At its peak, WildBlue added around 20,000 subscribers per month and gained a total of around 400,000 subscribers by the time it was sold to Viasat in 2009. More importantly, during 2021, Hughes and Viasat between them have lost almost as many subscribers in the U.S. as Starlink has gained. Hughes and Viasat assert that capacity constraints are a more important reason for this decline than competition, and Starlink claims that terminal production is currently constrained by chip shortages, but these factors are unlikely to be fully resolved before 2023. Until then, the jury will remain out on whether Starlink will be able to significantly expand the 1.7 million homes currently served via satellite broadband in the U.S. or whether competitors such as Viasat are correct and there is only limited room for further growth.

Today, the U.S. market accounts for the majority of the world’s roughly 2.5 million consumer satellite broadband subscribers, which is hardly surprising because U.S. consumers have a high willingness to pay for broadband (or put another way, terrestrial broadband is particularly expensive) and there is a large and relatively wealthy population in rural areas that are difficult to serve with wireline technologies. It is also hard to imagine a Starlink business plan where the majority of subscribers and revenues don’t come from the United States. But the continental U.S. is only 5% of the world’s land area and about 1.5% of the total surface area, so only a small fraction of Starlink capacity will be able to be directed towards serving U.S. customers. Hence a huge number of large, high-capacity Starlink version 2 satellites will be needed just to serve a few million subscribers in the U.S.

What is clear is that expectations for what Starlink can achieve in terms of closing the broadband gap in the U.S. must remain realistic. Fortunately, most of the commentators who made hyperbolic statements a couple of years ago that the U.S. should rely on Starlink instead of building more fiber have quieted down. However, some Wall Street firms still predict huge growth that supports a $100 billion-plus valuation for SpaceX, perhaps motivated by their desire to lead a future SpaceX IPO. To some degree, SpaceX itself was at pains to downplay expectations during 2021 and emphasize that Starlink will only be the best solution for the last few percent of users in rural areas. But that was also the route that Iridium took in the 1990s. When it became clear that satellite phones wouldn’t provide a realistic alternative to terrestrial cellular because of high costs and the inability to operate in most buildings, Iridium’s mantra became that it only needed 1% of the cellular market to be hugely successful. Today, it is far from clear that Starlink can achieve what it promised in winning bids to serve 600,000-plus homes in the FCC’s Rural Digital Opportunity Fund auction, without its much more ambitious version 2 constellation. And if the market for consumer satellite broadband does not grow dramatically, then that version 2 system — and the whole Starlink plan — may eventually crumble.

We will not know one way or the other about the ultimate size of this market during 2022, but with SpaceX representing the lynchpin of the NewSpace ecosystem, the risk has never been greater that we will ultimately see a repeat of the 1999-2001 crash in the satellite sector.

Tim Farrar is president of TMF Associates, a Menlo Park, California, consulting firm specializing in satellite communications and wireless spectrum.

This article originally appeared in the January 2022 issue of SpaceNews magazine.