SpaceX’s Starlink likely to lose in FCC’s $16 billion subsidy auction

14 min read


[responsivevoice_button rate=”1″ pitch=”1.2″ volume=”0.8″ voice=”US English Female” buttontext=”Story in Audio”]

SpaceX’s Starlink likely to lose in FCC’s $16 billion subsidy auction

  • SpaceX has launched hundreds of internet-beaming Starlink satellites and recently started a private beta for the network.
  • But the aerospace company, founded by Elon Musk, might try to defray some of its cost through Federal Communications Commission subsidies worth $20.4 billion.
  • Starlink performance tests published on Speedtest suggest that SpaceX might be eligible for the program, called the Rural Digital Opportunity Fund.
  • But industry experts say they doubt Starlink will get anything, because the fund strongly favors fiber-optic cables over satellite networks.
  • Still, a loss in the program’s $16 billion first phase might help SpaceX push for changes that allow it to better compete for a $4.4 billion second phase.
  • Visit Business Insider’s homepage for more stories.

SpaceX has a penchant for powering technological disruption with engineering savvy and federal assistance, most famously with its Falcon 9 rocket and Dragon spaceship programs.

But its next big play to glean some government dollars — to power a network of internet-beaming satellites — is likely to fail, industry experts say.

On Tuesday, SpaceX launched a new batch of desk-size Starlink satellites to orbit, boosting its operational fleet to about 600. The goal of the project, currently backed by private money, is to provide affordable high-speed web service across America and the rest of the world.

The company’s founder, Elon Musk (now the fourth-wealthiest person alive), told Business Insider in 2019 that it would take about 1,000 spacecraft to make the system “economically viable.” But that’s just the starting point: One day, as many as 42,000 Starlink satellites could simultaneously circle the planet.

spacex starlink internet satellite spacecraft solar panels arrays earth orbit illustration 00002

SpaceX may launch a constellation of up to 42,000 Starlink satellites around Earth.


SpaceX might have to make that happen almost entirely on its own. Industry experts told Business Insider the company will struggle to win a penny of a $20.4 billion subsidy program called the Rural Digital Opportunity Fund, which is managed by the Federal Communications Commission. The FCC plans to award companies that money in exchange for building up as much broadband access in as many underserved areas as possible.

The rules of the program stack the deck against approaches that don’t involve fiber-optic cables, and recent leaks of Starlink performance data from a private beta test almost certainly won’t help SpaceX.

“My guess, looking at the numbers, is the satellites don’t win anything,” Blair Levin, a telecommunications policy analyst, nonresident fellow at the Brookings Institution, and former FCC chief of staff, told Business Insider.

Applications for the first batch of that money, totaling $16 billion, were due July 15, and it’s unclear whether SpaceX submitted one. The company did not respond to a request for comment, and an FCC representative told Business Insider only that “the list of applicants has not yet been published.”

Given the company’s success in lobbying for the right to compete for these funds, SpaceX likely put itself in the running. And while a loss would sting, a rejection might become useful down the road.

“When you’re playing the long game, you understand that you win some, you lose some,” said Levin, who was a key figure in realizing the US National Broadband Plan (which got tens of millions of Americans online). “Every time you lose, you create an argument for why you should win the next one.”

A $20.4 billion push to help close America’s digital divide

us flag cowboy hat rancher farmer rural broadband phone GettyImages 1156602036

Rural populations in the US most poignantly feel the pain of unreliable, low-speed, or unavailable internet access.

Jerry Holt/Star Tribune via Getty Images

Those who live in remote areas of America know the problem well: The internet access is crummy, if it’s accessible at all.

“Without access to broadband, rural Americans cannot participate in the digital economy or take advantage of the opportunities broadband brings for better education, healthcare, and civic and social engagement,” the FCC said in January. 

The agency reported in 2017 that about 21 million Americans didn’t have access to broadband internet, a connection the commission defines as a minimum download speed of 25 megabits per second and upload speed of 3 Mbps. (The regulatory agency considers that fast enough to telecommute, do schoolwork, or stream 4K video, for example.) 

But as Business Insider’s Tyler Sonnemaker reported in March, that number is a vast underestimation; it’s more likely that 42 million to 162 million people in the US (as much as 50% of the population) don’t have broadband.

Over the past decade, the FCC has fueled a lot of infrastructure development to change that. In January, it announced the details of its latest subsidy program, the Rural Digital Opportunity Fund, paid for by fees tacked on to phone bills.

Phase one of the program, detailed in June, puts $16 billion up for grabs in a multiple-round reverse auction on October 29. Providers that can offer the highest internet speeds, the quickest latency (that is, round-trip signal delay), and the lowest infrastructure cost are most likely to get subsidies. Phase two, whose auction date and rules haven’t been determined, will offer another $4.4 billion to fill as many remaining gaps in rural internet service as possible.

SpaceX believes it can serve up affordable high-speed service to many of those Americans. The FCC appears to agree, since it has authorized thousands of Starlink satellites to launch through 2027.

“Satellite technology can help reach Americans who live in rural or hard-to-serve places where fiber optic cables and cell towers do not reach,” the commission’s chairman, Ajit Pai, said in a statement in February 2018 ahead of the approval.

That SpaceX would chase these subsidies to defray the costs of Starlink, even as it privately raises $1 billion to $2 billion a year, makes sense: Building, launching, and operating thousands of satellites is not cheap. In May 2018, Gwynne Shotwell, SpaceX’s chief operating officer and president, said Starlink might be “the most challenging project we’ve undertaken” and could cost “about $10 billion or more to deploy.” (Amazon also gave a $10 billion estimate for Kuiper, its planned and competing fleet of 3,236 internet satellites.)

“I think they can do what they want to without a penny of government money, and so therefore every penny of government money helps cushion their overall challenge,” Ernesto Falcon, a senior legislative counsel at the Electronic Frontier Foundation who focuses on broadband access and competition policies, told Business Insider. “If the government’s going to give you the ability to apply for free money, why wouldn’t you just take it?”

Starlink and other low-Earth-orbit fleets seem tailor-made to claim the new subsidies, since their coverage would reach almost anywhere on Earth and far exceed the speeds of yesteryear’s satellites.

But the FCC’s program appears to have left them in a lurch — and by design.

The FCC holds ‘serious doubts’ about satellite networks

space starlink satellite internet coverage animation signal cones_slow.2019 10 18 14_13_35

An animation of SpaceX’s Starlink satellite constellation providing internet coverage to the Americas.


Within the paragraph that permits satellite providers to compete for the new funding — a change SpaceX fought for — the FCC said it had “serious doubts that any low earth orbit networks” would qualify to compete in the fund’s most appealing category of subsidies, called the “gigabit” performance tier (meaning 1,000 Mbps).

“The purpose of the Rural Digital Opportunity Fund is to ensure that Americans have access to broadband, no matter where they live. It is not a technology incubator to fund untested technologies. And we will not allow taxpayer funding to be wasted,” Pai said in a statement accompanying the release of the phase-one rules.

Pai added: “A new technology may sound good in theory and look great on paper. But this multi-billion-dollar broadband program will require ‘t’s’ to be crossed — not fingers. So any such application will be given very close scrutiny.”

Notably, the final rules and Pai’s statements came less than two weeks after SpaceX submitted confidential data to the commission regarding Starlink’s “network latency testing and performance.”

The public now has some sense of those metrics via Reddit, after Starlink beta users (likely violating their nondisclosure agreements with SpaceX) tested their connections with Speedtest, which publicly posts anonymized results. Speedtest, run by a company called Ookla, helps people gauge their download speed, upload speed, and ping (a rough measure of latency) by shuttling data sent to and from a nearby server.

The results showed download speeds ranging from 35-60 Mbps and upload speeds ranging from 5-18 Mbps.

“The tests you link to all appear to be legitimate,” an Ookla representative, Adriane Blum, told Business Insider of the posted results. She added that Ookla had “robust” quality-assurance measures in place and vetted internet-service-provider names — in this case, SpaceX’s Starlink network.

Shrihari Pandit, a photonics engineer and CEO of Stealth Communications, which provides internet access to New York City businesses and researches new internet infrastructure technologies, also said the results appeared valid.

“There’s no way to really improve those tests locally,” he told Business Insider. He said that while it was possible to make an internet connection on Speedtest look slower or laggier than it is, that seems unlikely in this case.

spacex starlink satellite internet ookla speedtest speed test results bandwidth upload download ping july august 2020

A sample of anonymous Speedtest results for SpaceX’s Starlink satellite-internet beta from late July through early August 2020. via Reddit

In its 2016 FCC application to launch Starlink, SpaceX said the network could eventually provide up to 1,000 Mbps of bandwidth to an individual subscriber after full deployment.

Pandit said it wasn’t too surprising to see performance that’s about 4% as fast, since SpaceX is still building out, testing, and adjusting its network, and has upgrades planned for it.

“It’s kind of concerning that you’re getting these data rates so early in the game, though,” Pandit said. “I would suspect it should be a little higher than where it is, let’s say a couple hundred megabits per second.”

Still, at face value — and without knowing how many users are connecting to a satellite flying overhead at a given time — the results suggest SpaceX should qualify for phase one’s lower performance tiers. Called “baseline” and “above baseline,” these two levels dictate connections greater than 25 Mbps download/3 Mbps upload and greater than 50 Mbps download/5 Mbps upload, respectively.

Adding more Starlink satellites would open up more bandwidth to users, which would improve download and upload speeds — perhaps pushing Starlink into the program’s third tier, called “above baseline,” or connections better than 100 Mbps download/20 Mbps upload.

Pandit said SpaceX might also make speed improvements by updating software on satellites and user terminals, whose appearance Musk has described as a “UFO on a stick.” Specifically, he said Starlink could be tuned to better account for issues like signal loss due to wireless data passing through Earth’s atmosphere.

But SpaceX is, for now, shut out of the coveted “gigabit” tier, which calls for connections of 1,000 Mbps download/500 Mbps upload, and which the competition rules greatly favor in terms of making it easier to win bids. If there are many of those providers vying for subsidies across the US, as Levin expects there to be in the upcoming auction, little if any cash may be left over for SpaceX and other lower-tier competitors — if they’re even deemed eligible to bid in the first place.

One additional issue Pandit said he saw in the presumable Starlink performance data was the ping. The FCC auction rules heavily penalize a company’s bid if its service’s latency is higher than 100 milliseconds. The Speedtest results suggested that Starlink’s latency sits below that threshold, with pings reaching as high as 94 milliseconds. Yet Pandit noted the data appeared to come from relatively local connections, typically from rural Washington state (where SpaceX is beta-testing Starlink) to Los Angeles (where SpaceX is apparently basing network test servers).

A lot of Starlink user traffic would be served by local servers and should not, in theory, exceed 100 milliseconds. However, given a video call from rural Washington to New York City, Starlink in its current form — without satellites beaming data through space to each other via lasers, a planned feature — the latency would exceed 100 milliseconds, Pandit said. This is because the signal would have to hop onto a transcontinental fiber network, which might add about 60 milliseconds of delay.

“This would definitely create a problem for SpaceX,” Pandit said. “In order for them to resolve it, they’ll need to get additional satellites up in space.”

Pandit also said adding more strategically positioned ground stations would help. So would implementing Starlink’s planned laser-interlink technology, which would shuttle data at the speed of light through space (nearly twice as fast as light can travel through fiber-optic glass) from one satellite to another.

The test results are ‘not good,’ but they might not matter

FILE PHOTO: Ajit Pai, chairman of the Federal Communications Commission, speaks at the WSJTECH Live conference in Laguna Beach, California, U.S., October 21, 2019. REUTERS/Mike Blake

Ajit Pai, chair of the FCC, has vowed to closely scrutinize any satellite-based applications for the RDOF.


Despite the nascency of Starlink, Falcon said it’s hard to avoid comparing the Speedtest results, even for a beta test, to the marks SpaceX says it’s supposed to achieve. “So much of the rhetoric around SpaceX in the early years was gigabit or fiber-like speeds,” he added. “This is clearly not that.” 

In the context of FCC politics and the prospects of bidding for these subsidies, he added “it’s not good” to see such performance results.

Falcon said that “if you’re already leery of anyone saying ‘a satellite broadband option is great for you’ and the first public reporting from the press is ‘here are these pretty slow speeds,'” it wouldn’t be hard to make the leap to wondering why the government would pay for satellites instead of fiber.

But any influence from the leaked performance results might matter less than the official rules. Levin said the application and bidding process appeared to favor the digging of fiber-optic cables, not the launching of satellites. He said that was in part because another satellite-internet provider earned $122.5 million of $1.488 billion in a similar FCC subsidy auction called the Connect America Fund Phase 2 in 2018.

For a program that seemed geared toward building out wired connections, the win shocked some, apparently including people in areas that were set to receive the subsidized internet access.

“In the first half of the auction, the satellite guys won a significant portion of the bids, and there was criticism, including from the communities, that they had been locked into a non-future-proof network,” Levin said. “And so the FCC for the next auction adopted a formula that gives greater weight to the faster-speed, lower-latency networks.”

fiber optic cable light speed data rainbow shutterstock_348640364

Fiber-optic cabling.


Generally, the FCC’s view is that while burying fiber-optic cables may be tedious and expensive, it is ultimately the most durable method of getting Americans online.

“Fiber has a lifespan of 20 to 30 years,” Pandit said. Once cable conduits are laid, he said, replacing or upgrading them is fairly easy and inexpensive. Low-Earth-orbit satellites, by contrast, last just five to seven years. Musk has said as much, saying it’s a chance to also upgrade the spacecraft. 

The satellites also have to be carefully monitored to avoid collisions in orbit and to prevent the creation of space debris. That’s a problem ground cables don’t have.

Fiber has the capacity to shuttle the extreme amounts of internet traffic that are anticipated in the future. Meanwhile, satellite constellations like Starlink have a lower ceiling for bandwidth than fiber, because of both wireless-spectrum allocation and the physics of interference.

“If you didn’t put the fiber in the ground as the first step, you will have to do it later if all of the trends of internet applications and services continue,” Falcon said. “They are always going to require more and more bandwidth.”

That’s not to say SpaceX won’t win anything for Starlink through the FCC’s rural-broadband program. Levin conceded that he could be “totally wrong” because of the mind-numbing complexity of the reverse auction process (which he said has made some game theorists into multimillionaires).

“It depends on who else shows up,” he said. “You can say that the better baseball team will win, but sometimes they don’t. That’s why you’ve got to play the game.”

Even if SpaceX doesn’t score in the first few figurative innings, Levin thinks the company — especially as its network capacity and capabilities grow — could use its defeat to rally the phase-two rules in its favor. And industry experts think that, subsidies or not, Starlink could still become a highly successful and useful business.

“In the US, we built this interstate highway system to allow passengers and trucks to do commerce and allow people to travel between state lines. But we haven’t built an interstate highway system for fiber-optic cables, and that needs to happen,” Pandit said. “And until we do that, we will rely on companies like SpaceX to deliver a service reliably and more effectively to a wider range of people, where fiber simply could take decades to get deployed.”

You May Also Like