Recently, Elon Musk's SpaceX released its first earnings report since going public. During the highly anticipated earnings call, in addition to discussing rocket launches, satellite deployments, and AI computing power, SpaceX executives systematically disclosed, for the first time, Starlink's future expansion plans for mobile communications: the company plans to build a ground network based on small cell base stations across the United States, combining it with existing satellite D2D (Direct-to-Device) services to challenge major U.S. operators like AT&T, Verizon, and T-Mobile.

This announcement has sparked widespread debate across the global telecommunications industry. Is a company long regarded as a 'satellite internet service provider' now encroaching on the turf of traditional mobile operators? Is SpaceX truly set to become America's 'fourth major operator'?

If this development is seen merely as a 'satellite company crossing over to compete in the operator market,' it may underestimate Musk's ambitions. In fact, from Starlink's early focus on broadband coverage in remote areas to its current layout (layout) of D2D and ground-based small cells, and now targeting future 'physical AI' scenarios like automotive and robotics, SpaceX is attempting to redefine next-generation connectivity infrastructure. As satellites, cellular networks, Wi-Fi, and edge AI devices gradually converge, will traditional operators continue to dominate mobile communications? Is Starlink building a new mobile network or creating a global connectivity platform that transcends traditional telecommunications systems? This may be the true focus of this development.

Is Starlink Vastly Underestimated?

Based on currently disclosed information, Starlink plans to launch next-generation D2D satellites by 2027 and increase deployment of ground-based small cell base stations to break the traditional cellular cost model and create a hybrid communication service competitive with AT&T, T-Mobile, and Verizon.

A prerequisite for achieving all this is essential spectrum resources—after acquiring EchoStar's 65 MHz AWS-3/4 spectrum by the end of 2025, Starlink's spectrum capacity will increase from the current 5 MHz obtained through carrier partnerships to 70 MHz, a 13-fold growth. Benefiting from this, the number of future Starlink D2D satellites will increase tenfold, and these changes will enable its mobile communication services to reach '100 times' current levels by the end of 2027.

In terms of network construction, the company aims to avoid the massive capital expenditures typically required by large telecom operators. Instead of deploying traditional macro base stations, Starlink is exploring a more 'capital-efficient' approach—leveraging existing broadband satellite terminals to add small cell base stations to its infrastructure.

Musk commented, 'Compared to deploying large, extremely costly cellular base stations with difficult site selection, we believe we can confidently deploy a large number of small sites. These devices are essentially Starlink antennas that also provide mobile frequency connectivity, and they can be widely distributed. Currently, Starlink antennas are already installed on the roofs of numerous residential and commercial buildings, which typically offer excellent line-of-sight conditions for directly connecting to ground-based mobile devices. In fact, we believe this approach may provide better connectivity and higher bandwidth than services currently offered by cellular operators.'

SpaceX Chief Operating Officer Gwynne Shotwell confidently stated, 'We can capture a significant number of customers from these large operators because I believe our service will be superior.'

That said, this is not actually the direction Musk and his team are truly pursuing—they are not simply aiming to become America's fourth mobile operator, challenging traditional telecom companies solely through smartphone plans. More ambitiously, Musk and Shotwell emphasized in the earnings call their focus on connecting vehicles, robots, and other machines belonging to the 'physical AI ecosystem.'

Musk said, 'At some point in the future, it's not impossible for Starlink to provide the majority of global internet services—at least in countries where we're allowed to operate. This isn't a distant future but could happen in less than a decade. With the development of AI, humanoid robots, and vehicular robots, bandwidth demand will far exceed past levels. You might imagine that a person generates or consumes only a few hundred bits per second of data output daily. In fact, calculated over 86,400 seconds in a day, the average human output is even below 1 bit per second. Yet a computer can continuously generate billions of bits per second over 24 hours. That's why, with the advancement of AI and robotics, I believe bandwidth demand will surge dramatically. In fact, I think Starlink may be the only network truly capable of meeting this bandwidth demand.'

However, even if future demand for physical AI connectivity grows as rapidly as predicted, Starlink believes it can still withstand the inevitable decline in unit traffic pricing. Musk noted, 'Even if our per-bit data monetization capability drops tenfold, it still means Starlink's revenue could grow tenfold. I think people are severely underestimating Starlink.'

Is This Just a Wild Dream?

Since the news broke, various opinions have emerged, including some intriguing viewpoints.

Some view this as a wild dream—Dimitris Mavrakis, Senior Research Director at ABI Research, stated on social media, 'If Elon Musk believes that 5G small cells using Ver Starlink backhaul can provide better and cheaper services than AT&T,izon, and T-Mobile, he's indeed dreaming. Of course, if SpaceX/Starlink can deploy millions of such devices, the situation might change. But connecting these devices isn't the biggest issue—the real challenges are powering them and finding optimal deployment locations, not to mention business models, ROI, etc. Unless it considers leveraging existing Starlink sites, whose locations don't necessarily align with mobile coverage needs.'

Others have enthusiastically crunched the numbers for Musk—BNP Paribas previously estimated that building a nationwide 5G network in the U.S. would require approximately $100 billion in investment, based on constructing 50,000 to 70,000 sites; if C-band deployment needs are added, the site count could rise to 120,000. However, Richard Haas of Britain's LS Telcom believes the task could be accomplished for around $12 billion. He said, 'A macro base station might cost between $100,000 and $250,000, depending on complexity. To become a full-fledged mobile network operator (MNO), you might need 30,000 to 50,000 sites. Calculating based on the highest estimates, the total cost would be around $12 billion. For a company like SpaceX, which spent nearly four times that amount just on acquiring EchoStar's spectrum, this is entirely feasible.'

Nevertheless, capital is clearly not the core issue facing SpaceX. This seems to present a binary choice: either Starlink builds a fourth mobile network in the U.S., or it operates as an MVNO (Mobile Virtual Network Operator) through partnerships with major telecom carriers.

However, the second option now seems highly unlikely, as operators—whose positions are being challenged and threatened—have fully blocked Starlink's MVNO access. Verizon CEO Dan Schulman took the firmest stance, explicitly stating that Verizon would not provide any 'backdoor' for Starlink to enter MVNO operations, nor would it allow Starlink to access its MVNO system. He also emphasized that Verizon's future satellite partnerships would primarily rely on the satellite communications D2D joint venture established this May with AT&T and T-Mobile, aiming to create a unified satellite D2D wholesale market through device standardization and spectrum sharing, without opening network resources to Starlink.

AT&T CEO John Stankey, meanwhile, dismissed the necessity of an MVNO partnership with Starlink from a commercial logic perspective. He stated that AT&T currently covers approximately 98% of mobile traffic, with the remaining 2% of coverage gaps to be resolved next year through D2D services in partnership with AST SpaceMobile, eliminating the need to address a non-existent problem through an MVNO agreement with Starlink.

This leaves Starlink with what appears to be the only path forward. In response, industry insider Bubley insightfully pointed out that the current framework for understanding SpaceX is entirely misguided—the real question isn't whether SpaceX can match AT&T or Verizon with a cheaper 5G system, but whether it can redefine the very meaning of 'mobile communications.' In his view, 'mobile' has already become an umbrella term covering various connectivity scenarios, including: fixed wireless access in homes and workplaces; 'nomadic' connectivity in stations, cafes, and stores; true mobile connectivity in vehicles and trains; and coverage of 'dead zones' in remote areas.

Currently, the first scenario is largely dominated by Wi-Fi, while satellite communications are beginning to address the last scenario. If Starlink can cover the intermediate connectivity needs through self-construction or acquisitions, traditional telecom operators will face significant challenges. Operators might then be reduced to selling traditional 'OG mobility' services rather than providing universal connectivity across all scenarios. Bubley believes this would transform the entire business model. In the future, users may no longer rely on a single mobile plan to cover all needs but instead combine different types of connectivity services—including macro cellular networks, D2D satellite communications, traditional Wi-Fi, and some form of nomadic access. These services might be transparently managed by device software or AI.

In this context, mobile operators would no longer be the absolute center of user connectivity but merely one component within a broader connectivity ecosystem. At that point, Starlink Mobile wouldn't need ubiquitous coverage to disrupt the entire value system.

References: Musk’s Starlink plots small cell rollout to rival US mobile operators——telecomtv; SpaceX targets US carriers with small-cell terrestrial rollout and mega demand from physical AI——RCR Wireless; Starlink Mobile – what is Musk actually building, beyond the chatter?——RCR Wireless; Breaking: Starlink Faces Collective Blockade——5G and 6G Massive Small Cells + Next-Gen Satellites: SpaceX Plans to Completely Disrupt Operators——C114