SpaceX Wants to Be Your Phone Company. It's Building on the Ruins of the Last Company That Tried.

Background: The Fourth Carrier That Didn’t Happen

The U.S. has had only three facilities-based national carriers since T-Mobile absorbed Sprint in 2020. Regulators approved that merger on the condition that a new competitor would replace Sprint.

That competitor was Dish Network. Dish took over Boost Mobile and its roughly 9.3 million subscribers, and built a greenfield 5G network that launched in Las Vegas in 2022.

It did not take hold. Boost had about 7 million customers by 2025, fewer than when Dish bought it. Analysts blamed marketing and sales more than the network itself.

In 2025, Dish’s parent EchoStar sold its airwaves. AT&T agreed to pay $23 billion for 3.45 GHz and 600 MHz licenses. SpaceX agreed to pay about $17 billion for AWS-4 and H-block licenses, then $2.6 billion in stock for unpaired AWS-3 licenses.

Boost moved all of its customers onto AT&T’s network on Nov. 15, 2025. A federal judge later released EchoStar from its obligation to build and run a national network.

In simple terms: Washington tried to create a fourth national phone network in 2019. The effort failed, and SpaceX now owns much of the spectrum it was built on.

How the Plan Fits Together

SpaceX is building a carrier in layers: satellites for coverage, mid-band spectrum for capacity, low-band spectrum to reach indoors, and ground sites to carry traffic where people actually are.

Satellites. SpaceX’s current direct-to-cell fleet is roughly 650 satellites using about 5 MHz of partner spectrum, mainly through T-Mobile’s T-Satellite service. In October the FCC authorized 15,000 next-generation direct-to-device satellites.

Mid-band spectrum. On May 12, 2026, the FCC approved SpaceX’s purchase of about 65 MHz from EchoStar, with waivers allowing terrestrial, satellite or hybrid use, tied to buildout obligations. Three Tristar AWS-3 licenses covering about 4% of the population bring the total to roughly 75 MHz. EchoStar’s most recent quarterly filing says final ownership transfer is expected on or about Nov. 30, 2027.

Low-band spectrum. On Oct. 8, SpaceX agreed to buy Grain Management’s nationwide 800 MHz portfolio — up to 14 MHz of paired spectrum. The price was not disclosed; The Wall Street Journal reported about $8 billion in cash. Grain had acquired the licenses from T-Mobile only in August. The deal needs FCC approval.

Ground sites. On SpaceX’s first earnings call in August, President Gwynne Shotwell described adding towers and small cells to the satellite network. The 80,000-site talks are the first sign of that at national scale.

In simple terms: low-band signals travel far and pass through walls. Mid-band carries more data over shorter distances. Satellites fill in where there are no towers. Until now, SpaceX had the satellites and some mid-band, but not the other two pieces.

Asset	Amount	Price	Status
EchoStar AWS-4 + H-block	50 MHz	~$17B	FCC-approved May 2026; transfer expected ~Nov. 2027
EchoStar unpaired AWS-3	15 MHz	$2.6B in stock	FCC-approved May 2026
Tristar unpaired AWS-3	3 licenses, ~4% of population	Not disclosed	FCC-approved October 2026
Grain 800 MHz	Up to 14 MHz paired	~$8B (WSJ)	Pending FCC
Next-gen D2D satellites	15,000	—	FCC-authorized October 2026
Cell sites	80,000+	—	In talks (Bloomberg)

How the Incumbents Have Responded

AT&T, Verizon and T-Mobile have closed the fastest door into their market. The quickest way for a newcomer to sell phone service is a wholesale lease on an existing network, known as an MVNO deal. All three CEOs have said they will not offer SpaceX one. Asked about it at an investor conference, Verizon CEO Dan Schulman answered with a single word: “no.”

The three also formed a joint venture that pools some of their spectrum for satellite providers to cover rural dead zones. Analysts at LightShed questioned whether the Justice Department would allow it, given that the same three companies had each refused SpaceX wholesale access.

The fight has also moved to the FCC. On Oct. 22, the industry group CTIA filed notice of meetings with advisers to all three commissioners, objecting to parts of a draft rule that would expand satellite-to-phone service from five spectrum bands to 13. CTIA asked the FCC to drop the 2.5 GHz and 3.45 GHz bands and to keep carriers’ consent and termination rights. The commission is scheduled to vote on Oct. 29.

The relationship is not purely adversarial. T-Mobile remains SpaceX’s U.S. direct-to-cell partner through T-Satellite, even as the two now compete for the same customers.

Potential Antitrust Issues

The antitrust questions run in both directions: whether the three incumbents are coordinating to keep SpaceX out, and whether SpaceX is becoming too dominant itself.

The carriers’ joint venture. The three largest carriers announced their satellite joint venture on May 14, 2026, two days after the FCC approved SpaceX’s EchoStar purchase. SpaceX’s vice president of satellite policy, David Goldman, publicly questioned whether the Justice Department would allow it. He cited LightShed Partners, whose analysts argued that the same three companies had each refused SpaceX wholesale access and were now pooling scarce spectrum just before SpaceX’s IPO.

The carriers describe the venture as a way to end rural dead zones and build a stronger wholesale satellite market. AT&T CEO John Stankey has said he expects a longer Justice Department review rather than a quick sign-off. New Street Research judged that the department would likely accept the carriers’ efficiency argument — unless SpaceX pushed for a serious investigation.

The public messaging has continued. In October, Tim Höttges, CEO of T-Mobile’s parent Deutsche Telekom, argued that carriers should not offer SpaceX a wholesale deal, saying it could hurt their own businesses.

What the law generally allows. U.S. antitrust law usually lets a single company refuse to do business with a rival; the Supreme Court said so in Verizon v. Trinko (2004), itself a telecom case. Parallel behavior by competitors is also not illegal on its own without evidence of an agreement, under Bell Atlantic v. Twombly (2007), another telecom case.

In simple terms: each carrier saying no to SpaceX separately is generally legal. A formal joint venture among all three is an actual agreement between competitors, so regulators can weigh whether its benefits outweigh any harm to competition. No public evidence so far ties the venture’s terms to how its members deal with SpaceX.

SpaceX’s own position. Critics have raised the opposite concern. In December 2025, Sen. Elizabeth Warren and Rep. Greg Casar urged the Justice Department and FCC to scrutinize the EchoStar purchase, arguing it could entrench SpaceX’s dominant position in satellite communications.

The FCC disagreed. In approving the EchoStar deal, it found SpaceX would hold at most 75 MHz in any local market, far below the 385 MHz level that triggers closer competition review, and said the deal could increase competition. SpaceX’s Grain application says its holdings would top out at 89 MHz, including 14 MHz below 1 GHz. No one petitioned to block the Tristar purchase.

The real test may come later. If SpaceX formally asks the Justice Department to investigate the venture, or if the venture’s final terms link spectrum pooling to wholesale access, the antitrust question would move from commentary to enforcement.

What It Means for Consumers

No one can switch to a Starlink Mobile plan on SpaceX’s own network soon. Its next-generation “5G-quality” satellite service is targeted for the first half of 2028. The EchoStar spectrum does not fully transfer until late 2027, and the Grain purchase still awaits FCC approval. SpaceX says most phones already support the 800 MHz band, so that layer would not require a new handset.

Prices are the central question. Wireless service prices rose 5.9% in August 2026, the largest one-month increase on record in Bureau of Labor Statistics data. All three carriers raised prices this year: AT&T by up to $20 a month on some legacy plans, T-Mobile about $6 a month for customers moved off grandfathered plans, and Verizon $5 for new Unlimited Ultimate customers.

The longer trend points the other way. The carriers’ trade group, CTIA, says inflation-adjusted unlimited-plan prices fell 10% over the past year. BLS also changed how it measures wireless prices starting with July 2025 data, so recent months are not directly comparable to the older series.

Would people switch? In an October survey of about 800 consumers by TD Cowen, 34% said they would consider switching to Starlink Mobile. Nearly a quarter said they were put off by Musk-affiliated products, and 12% doubted its performance. TD Cowen’s analysts concluded the service would likely have to win on price.

How customers rate the three carriers. We found no recent public favorability poll covering all three. The closest measures are satisfaction studies of each carrier’s own customers:

Carrier	ACSI 2026 (0100)	J.D. Power postpaid (01,000)	J.D. Power prepaid (01,000)	Would consider Starlink Mobile (TD Cowen, Oct. 2026)
T-Mobile	78	631	629	30%
Verizon	76	593	602	23%
AT&T	76	587	604	52%

Sources: ACSI 2026 study; J.D. Power Wireless Carrier Satisfaction scores (fielded JuneNovember 2025) as compiled by Clark.com; TD Cowen survey via Fierce Network.

ACSI found little separation among the three networks, even as industry satisfaction hit a record 77. The highest-rated wireless services were low-cost brands that resell the big three’s networks, at an average of 79. On brand image, TD Cowen’s October survey found Verizon had retaken the No. 1 spot from T-Mobile.

That pattern matters for SpaceX. If customers already see the networks as roughly equal, a newcomer would likely compete on price and service rather than coverage — and AT&T’s customers, the least satisfied in J.D. Power’s postpaid ranking, are the most open to switching.

The Phone Question: Operating Systems and Apps

Starlink Mobile is a network, not a phone. It is designed to run on the iPhones and Android phones people already own, so it sidesteps the app problem that sank Microsoft’s Windows Phone. Its real dependency is on Apple, Google and the chipmakers.

Why Windows Phone failed. Microsoft ended support for Windows 10 Mobile on Dec. 10, 2019, and recommended that customers move to iOS or Android. Developers never committed in large numbers, and the “app gap” with iOS and Android drove users away. Even Microsoft’s own Skype app took two years to reach the platform, and the company wrote off $7.6 billion on its Nokia phone business.

In simple terms: a phone is only as useful as its apps, and developers build for the platforms that already have the most users.

Why a carrier avoids that problem. A carrier does not need its own operating system. Starlink’s current satellite service works with existing LTE phones, and SpaceX says most phones already support its new 800 MHz band. Apps would run on iOS and Android exactly as they do today.

Where the dependency lies. SpaceX’s satellite spectrum is the harder part. Musk said in June that phones able to use the EchoStar spectrum will probably start shipping in around two years. That requires Apple, Samsung, Google and chipmakers to build support for those bands into new devices. Apple was reported in January to be in talks with SpaceX about supporting Starlink’s direct-to-cell service on iPhone 18 Pro models, according to The Information. A full carrier also needs device makers to certify its network for everyday features such as eSIM activation and Wi-Fi calling.

Would SpaceX build its own phone? Musk said in January that a Starlink phone was not out of the question, but that it would be built mainly to run AI models efficiently. After Reuters reported in February that SpaceX was considering one, Musk posted, “We are not developing a phone.” He also publicly denied later reports that SpaceX had shown investors a prototype. Deepwater’s Gene Munster predicted on Oct. 9 that SpaceX will eventually make its own handset anyway.

If it does, the Windows Phone lesson applies directly. A phone running Android would inherit Android’s app store; one running a new operating system would have to build an app store, developer tools and services from zero.

Bottom line: the operating-system risk applies only if SpaceX builds its own handset. The nearer-term risk is that its satellite service depends on Apple and Google adding support — and Apple already has its own satellite partner, Globalstar, for iPhone emergency messaging.

What It Means for Tower Owners

For tower owners, SpaceX is a potential new tenant. Any agreements could be worth billions of dollars, according to Bloomberg. Crown Castle CEO Chris Hillabrant did not name SpaceX on Wednesday’s earnings call, but told analysts, “You can assume that we’re talking to all parties.”

Access to 80,000 sites is not the same as building 80,000 base stations. SpaceX would still need power, backhaul, local maintenance and site-by-site agreements, and as a carrier it would take on licensing, universal service and roaming obligations.

What It Means for SpaceX Shareholders

Starlink Mobile is being funded by the same Starlink cash flows that support SpaceX’s $2 trillion valuation. Shareholders are weighing whether it becomes a new revenue line or a third major drain on capital alongside AI and Starship.

Where the stock stands. SpaceX has traded on Nasdaq since June 12. Shares opened at $150 and were at $147.50 on Oct. 20, a market value of about $2 trillion, before third-quarter results. They closed at $159.31 the next day. The Bloomberg tower report added more than 2% in premarket trading Friday.

Starlink pays the bills. Third-quarter revenue rose 161% to $14.58 billion, and SpaceX reported net income of $4.4 billion. Connectivity, mostly Starlink, brought in $5.25 billion, up 69%, and CNBC noted that almost all of the company’s profit still comes from Starlink. Consumer subscribers reached 13.6 million.

The spending is the main concern. Third-quarter capital spending was $19.59 billion, $16.44 billion of it for AI infrastructure. SpaceX guided fourth-quarter capex to $35 billion to $40 billion. On top of that sits roughly $28 billion in announced or reported spectrum deals. TD Cowen estimates a full ground network would cost more than $80 billion. Shotwell has declined to give a figure, saying the company has new ideas that would be capital efficient.

The size of the prize is disputed. SpaceX’s IPO materials assigned Starlink Mobile a $740 billion addressable market. GSMA Intelligence estimated the realistic figure at about $15 billion a year. On the bullish side, Deepwater’s Gene Munster said a SpaceX phone on its own network could undercut carriers on price, and William Blair expects a combined broadband-and-mobile package.

The spectrum has resale value. If the carrier plan stalls, the airwaves do not become worthless. EchoStar raised more than $40 billion selling its licenses after its own network effort failed, and Grain resold the 800 MHz portfolio two months after acquiring it. Spectrum could also be traded for the wholesale deal the incumbents have so far refused — though there is no guarantee SpaceX would recover what it paid.

This article is for informational purposes and is not investment advice.

Analysis: A Carrier, or a Bargaining Chip?

The central question is whether SpaceX intends to run a fourth network or to accumulate enough assets to force its way into someone else’s.

The case that it is building. SpaceX’s own executives have been explicit. On the August earnings call, Musk said spectrum and satellite launches will pave the way for Starlink Mobile to become a major mobile carrier, and Shotwell said she expects to win “quite a few” of the big three’s customers. Its spectrum now spans low-band and mid-band, with FCC waivers for ground use. And tower leases are a step toward running a network, not just negotiating for one.

The case that it is leverage. MoffettNathanson argued in July that the likeliest path is for SpaceX to collect tradeable assets — mostly spectrum — to pry open an MVNO deal, the route cable companies Comcast and Charter used with Verizon. BNP Paribas analyst Sam McHugh called a seamless satellite-plus-small-cell network “far-fetched”. Morgan Stanley’s Sean Diffley said the perceived risk to carriers exceeds the actual risk over the next one to two years.

The physics argument favors the incumbents. Ground networks get their capacity by reusing the same spectrum across dense grids of cells. A satellite beam covers far more area with far less capacity per person, which is why analysts treat satellite service as a supplement in cities, not a replacement.

What Dish’s failure does and doesn’t predict. Dish held more spectrum than SpaceX does today and had built a working network. It still lost customers. The bottleneck was not airwaves; it was distribution, devices, retail and customer acquisition — the expensive, unglamorous parts of a carrier.

SpaceX enters with advantages Dish lacked: a consumer brand, an existing paying customer base, public-market access to capital, and a satellite layer no rival owns. It also enters with a constraint Dish did not have — the same capital is competing with Starship, the satellite constellations and AI infrastructure.

Bottom line: SpaceX has bought the pieces a fourth carrier needs. Whether it assembles them, or uses them to negotiate a better deal from the three it is challenging, is still an open question.

Could Musk Remake Wireless? A Look at the Playbook

Musk’s track record is the main reason investors and incumbents take Starlink Mobile seriously. SpaceX entered launch and satellite internet as an outsider and reset the economics of both. This section looks at how that playbook might apply to wireless — and where it might not. It is analysis, not a prediction.

Where the playbook worked. Reusable rockets gave SpaceX what research firm Sacra estimates is a 5 to 10 times lower cost per ton than competitors. In 2025, SpaceX flew 165 of the 192 U.S. orbital launches, about 86%. Starlink went from no customers to 13.6 million consumer subscribers and now supplies almost all of SpaceX’s profit.

Where it did not. Not every Musk idea has transformed its industry. Hyperloop One, a startup built on Musk’s 2013 high-speed tube concept, shut down in 2023 without carrying a passenger after spending more than $450 million. SpaceX tore down its own Hyperloop test track in 2022 to make room for parking.

Five ways Musk could try to change wireless

One plan that works everywhere. SpaceX CFO Bret Johnsen has said that eliminating dead zones and offering native global roaming are the pillars of Starlink Mobile, and the FCC has granted SpaceX authority to offer telecom service beyond the U.S. A single plan with no roaming charges would attack a fee travelers know well.
Owning the whole supply chain. SpaceX builds its own satellites, launches them on its own rockets, and designs its own chips for the next generation of mobile satellites, according to Starlink. No U.S. carrier controls its network costs that way. That vertical integration is what cut launch prices.
Simpler, cheaper pricing. TD Cowen expects Starlink Mobile would have to win on price, and Deepwater’s Gene Munster has said a SpaceX network could undercut the carriers. SpaceX already sells Starlink online without retail stores, much as Tesla sells cars without dealerships.
Bundles no carrier can match. A single account could combine Starlink home internet, mobile service and satellite coverage. SpaceX’s merger with xAI also gives it its own AI models, which could be built into the service.
Forcing change without winning. Even if Starlink Mobile never wins large market share, the threat has already moved the incumbents: they formed a joint venture, and analysts have cut their price targets. A credible fourth competitor can change prices and terms before it signs its first customer.

Why wireless may be harder

Launch was a market where SpaceX was its own biggest customer: about three-quarters of its 2025 launches carried Starlink satellites. In wireless, SpaceX must win customers away from three companies that already compete on price, hold far more spectrum, and score near record highs on customer satisfaction.

The constraints are also different. Musk’s main lever has been driving down cost, but the binding limits in wireless are spectrum capacity in dense cities, FCC and Justice Department oversight, and the cooperation of Apple and Google on devices.

Bottom line: the question is not whether Musk can disrupt an industry — he has, in launch and satellite internet. It is whether wireless rewards the things he is best at, or the things he has the least control over.

Conclusion

In 2019, regulators tried to manufacture a fourth national carrier and got a company that could not keep its customers. Seven years later, a different company is assembling the same kind of network from the remains of that attempt — this time without a government mandate behind it.

The next few months will show which way SpaceX is heading. Signed tower leases would point toward a real build. A sudden wholesale deal with one of the three incumbents would suggest the spectrum was the price of admission all along.

Key Takeaways

SpaceX is in talks for access to at least 80,000 U.S. cell sites, with Crown Castle among the tower owners named by Bloomberg.
It has agreed to spend roughly $28 billion on EchoStar and Grain spectrum, covering both low-band and mid-band frequencies.
AT&T, Verizon and T-Mobile have all refused to lease SpaceX wholesale access and formed their own satellite joint venture.
SpaceX’s next-generation consumer service is not targeted until the first half of 2028.
The carriers’ joint venture faces possible Justice Department review, while critics in Congress have questioned SpaceX’s own growing dominance; the FCC has so far found no spectrum-concentration problem.
Starlink Mobile is designed to run on existing iPhones and Android phones, avoiding Windows Phone’s app problem; Musk has said SpaceX is not developing its own phone, though phones that use its satellite spectrum are about two years out.
Customers rate the three carriers closely: ACSI scores range from 76 to 78, with T-Mobile on top. AT&T customers are the most open to switching (52%).
For shareholders, Starlink funds the company’s profits, but a full ground network could cost more than $80 billion on top of record AI spending.
Musk’s playbook — vertical integration, lower costs, direct sales and global roaming — reset launch and satellite internet; wireless tests whether it works where spectrum, regulation and device makers set the limits.
Analysts are divided on whether SpaceX will build a full network or use its assets to negotiate access to an existing one.

What to Watch

Oct. 29: FCC vote on expanding satellite-to-phone spectrum, opposed in part by CTIA.
FCC review of the Grain 800 MHz purchase.
Whether the tower talks produce signed leases, and with whom.
SpaceX’s fourth-quarter results: whether capex lands in the $35 billion to $40 billion range and how much goes to mobile.
Whether Apple, Samsung and chipmakers announce support for SpaceX’s satellite spectrum in new phones.
Monthly BLS wireless price data, for any sign of carriers cutting prices in response.
Nov. 30, 2027: expected final transfer of the EchoStar spectrum.
Any change in the carriers’ refusal to offer SpaceX a wholesale deal.