A day (or two) after SpaceX talked about plans for a terrestrial Starlink mobile network, analysts have started to debate whether Musk is building a fourth US carrier, optimising satellite architecture, or redefining mobile connectivity altogether.
In sum – what to know.
Big build debate – Analysts are split over whether SpaceX can justify, even afford, a national terrestrial network build-out, with estimates ranging from $12 billion to $100 billion.
Different disruption – Others argue that Starlink does not need to replace telcos; it only needs to unbundle connectivity to leave the incumbent selling only ‘actual’ mobility.
Engineering ambition – Others suggest dish-mounted small cells may be about improving satellite efficiency rather than creating a fourth national mobile network, at least for now.
Everyone had time to digest SpaceX’s grand-standing yesterday (actually late Tuesday), spurred by a quarterly scorecard that said the firm almost-doubled its revenue and nearly twice-tripled (six times) its net loss in the period, versus a year ago, and confirmed what everyone already suspected, and also what they didn’t: that it will build a terrestrial Starlink network in the US, and that it will do so using small cells? Well, all the excitable ‘influencers’ on LinkedIn, liberated by AI to shout tabloid ‘insights’ into corporate echo-chambers, have had their say. Everyone sitting comfortably? Well then, let us consider the sharpest voices among them – which have discerned the nuances in this sector for years. Their responses are telling, and worth re-telling.
At press yesterday (July 5), RCR cobbled together another half-and-half op-ed / leader column on top of the newsletter that put focus on comments from equity analyst BNP Paribas, which was quick to the draw following Starlink’s results on Tuesday (July 4) evening. It went alongside a straighter report that picked up the main news lines from its earnings call, including about building a telco-beating (“better”) service via a distributed network of small cells, plus hot-takes from analysts, still getting their heads around it. “Elon Musk is nuts if he thinks 5G small cells with Starlink backhaul can offer better (and cheaper) service than AT&T, Verizon, and T-Mobile,” retorted Dimitris Mavrakis at ABI Research. But it was the BNP Paribas viewpoint that we finished on.
Sam McHugh, head of telecom, cable and satellite research at BNP Paribas, reflected: “We continue to believe [SpaceX has] a weak hand in trying to generate meaningful revenue / returns from wireless in the US, but it also seems clear it intends to keep talking up its mobile opportunity – presumably with an eye to getting one of the carriers to break and offer an MVNO deal… This overhang is likely to persist through to the upper C band auction in 2027, unless SpaceX comes to some wholesale D2D agreement… that puts to bed this risk.” He conflated the talk about “femtocells” on broadband dishes with “the kind of Wi-Fi offload solution discussed with Charter”, and suggested a standalone version, as pitched, would be “even more subscale / inferior” than a deal with Charter.
Starlink build strategy
Since then, everyone has had a say, sometimes interesting. Richard Haas at LS Telcom in the UK says the cap-ex is not the problem. (Clearly so, for Musk; but anyway…) BNP Paribas puts the cost of a national US 5G build at about $100 billion, based on 50,000-70,000 sites, rising to 120,000 with the C-band. Haas reckons it can be done for about $12 billion. He said: “A macro base station could cost $100,000-$250,000, depending on complexity. To be a fully-fledged MNO, you may need 30,000-50,000 sites. That brings the cost, assuming the high end of both estimates, to $12 billion… For SpaceX, a company that spent nearly four times that much on EchoStar’s spectrum, that is completely doable.” It is not that a network is cheap, but it is not that much, and capital is hardly an issue.
It looks like a binary choice: either Starlink builds a fourth network in the US or piggyback on the big telcos as an MVNO. Which is not to say rumours about a Verizon or T-Mobile acquisition do not still cast a shadow – like the company’s own strategy, as stated barely 24 hours before, has been rejected out of hand. But Kyung Mun at Mobile Experts takes the SpaceX chiefs on their word, and maps the build option. Starlink has three-to-four million Starlink connections in the US, so that’s a baseline of future small cells [for a] Starlink network on the ground (rooftops).” Certainly, this follows the claims on the call about more cost-efficient rollout compared to standard telco networks. Mun raises the prospect of SpaceX using the CBRS band, even, or buying some of the C-band; he also points to likely deals with Apple and Samsung, and “CPE/small cell ‘combo’ devices” in urban areas.
Mun is not the only one to back the build strategy. Trevor McLaughlin, a consultant at Autonomous Systems, wrote: “I am fascinated by all of the posts regarding SpaceX not being able to offer a Starlink Mobile service without an MNO/MVNO relationship. I don’t find it too far outside of the realm of possibility that a version of the cable-co playbook could work… Ninety percent of Comcast Xfinity’s mobile traffic goes on Wi-Fi; [the rest] goes via its MVNO with Verizon.” Interestingly, McLaughlin also raises the unconventional idea that SpaceX could give spectrum to enterprises to “build and deploy private 5G networks” on the proviso they connect passing Starlink customers. The idea is to turn enterprise investments into public infrastructure, effectively. Hmm; but it is out of the box, at least.
Starlink master strategy
The sharpest reads are probably from Dean Bubley and Vish Nandlall, both familiars in these pages. Bubley says the framing is all-wrong. The question is not whether SpaceX can match AT&T or Verizon with a cheaper 5G system, but whether it can redefine what mobile even means. (Which is what RCR wrote yesterday.) The way he puts it, ‘mobile’ is an umbrella term for different connectivity scenarios, already: fixed wireless at home and work; ‘nomadic’ access in stations and cafés and shops; proper mobility on the go, in cars and trains; and ‘dead-zone’ coverage in far-away places. Wi-Fi has the first category sewn up, he says; satellite is starting to address the last. If Starlink can absorb some of the middle layer – whether it builds or buys – then the telcos have a problem.
They are left selling only OG mobility – rather than general-purpose connectivity. That, Bubley suggests, changes the commercial equation. Instead of one mobile subscription covering every scenario, users could eventually piece together different connectivity services – macro cellular, D2D satellite, standard Wi-Fi, and some kind of nomadic access – managed transparently by device software or AI. In that world, mobile operators become just one ingredient in a broader connectivity stack, rather than the dominant retail relationship. “Starlink Mobile doesn’t need to be ubiquitous in order to be disruptive to the overall value proposition,” he writes.
Meanwhile, Nandlall questions a different assumption: that putting small cells on Starlink terminals is necessarily the start of a fourth national mobile network anyway. He argues the architecture makes more sense as a way to improve the efficiency of Starlink’s satellite system itself – which tallies with the “100-times better” statement on the call. A rooftop terminal has vastly more antenna gain and link budget than a handset, allowing traffic to be shifted away from scarce D2D spectrum onto Starlink’s much larger Ku- and Ka-band broadband payloads. “The dish moves traffic off the scarce direct-to-cell payload and onto the much larger broadband payload,” he writes. “Starlink is being good-engineers, but there is a clear bottleneck that separates it from being a national carrier.”
The terrestrial challenge is a stubbornly terrestrial one, he says. Even with rooftop small cells, a meaningful urban network still requires hundreds of sites across every city, each needing to be acquired, permitted, powered and maintained. “The dish does not replace the terrestrial network,” he writes. The real signs that SpaceX intends to become a fourth US carrier, he argues, would be national tower agreements, a large-scale urban site acquisition programme, and aggressive bids for more terrestrial spectrum. Back to where we were then. Until then, the easiest reading is that SpaceX is making its satellite network more capable, while keeping pressure on incumbent operators over potential MVNO and wholesale deals. Like McHugh says at the top of the piece.
