Much of the talk at yesterday’s Satellite Innovation conference in Mountain View, including my presentation, focused on the challenges that SpaceX poses for the rest of the space industry and (more importantly) the wider telecom sector. In the background there’s the additional threat from the “missing pieces” that Musk highlighted during the SpaceX results call last week. That is widely recognized as SpaceX’s plan to acquire the rest of EchoStar in a share-based transaction, which would allow SpaceX to pick up the remnants of DISH Wireless network, along with the rest of EchoStar’s spectrum in the AWS-3 and CBRS bands, and save the cash outlay that would otherwise have been paid for the AWS-4 and Band 70 spectrum.

The benefit for EchoStar holders, and Charlie Ergen in particular, would be that a share-based transaction could be tax free, unlike the cash sales of spectrum. These direct negotiations with Charlie Ergen are the reasons that the tower companies have been so vocal in their recent results calls (for example about rights to abandoned equipment), and why we saw chatter about an 80,000 tower buildout last week (since a larger commitment by SpaceX could allow for a better price on the existing leases). And it would not be surprising if the deal was announced as soon as next week, given that mediation is scheduled through November 4, although Ergen is known for being a tough negotiator and might make it hard to reach an agreement on price.

But that’s mainly a telecom industry issue, and what was animating numerous people I spoke to yesterday, is the expectation that Amazon will soon announce the acquisition of Viasat, mainly for their international spectrum holdings, but also to combine Viasat’s Ka-band GEO business with Amazon Leo, and strengthen Amazon’s position in the government and aviation market (and incidentally remove Telesat’s biggest commercial customer for Lightspeed). Equatys would be abandoned as a satellite project, although Amazon would need to maintain a relationship with Space42 to coordinate access to the L-band spectrum in EMEA.

I’m also told AST has attempted to put together a rival bid, but that would require significant support from the telcos, and would be vehemently opposed by Viasat management. In contrast, selling Viasat to Amazon could be seen as a graceful exit, before Viasat’s current GEO business is overwhelmed by LEO competition.

In the D2D market, Amazon is seen as the most realistic potential savor by the telcos, as belief that AST will ever deliver a useful service ebbs away. The telco JV seems poised to sign MOUs with both Amazon and AST (since there’s no point in throwing AST under the bus at this point in time), although it remains to be seen how quickly Amazon would be able to bring a new D2D service to market (beyond Globalstar C3 which is presumably only going to work on iPhones) and how much capacity it would be able to provide in the US.

Clearly the FCC is also now taking the position that the entire LEO market will mostly just be a SpaceX vs Amazon fight, and yesterday approved Amazon’s acquisition of Globalstar, removing one of the possible holdups to a deal. What that means for AST’s deal with Ligado remains to be seen, and an FCC decision on the SkyTerra Next plan might wait until clarity emerges on Amazon-Viasat.

And AST’s relationship with Ligado remains tense: Ligado does not benefit at all from the high valuations now being put on MSS spectrum, because AST pays Ligado just enough to cover the ongoing lease payments to Viasat. Thus Ligado’s only objective is to settle its litigation with the US government and walk away with a large amount of money (or more likely auction bidding credits), giving up its spectrum rights in exchange, which would leave AST in the lurch (and perhaps return more spectrum to Viasat).

In this case RocketLab and Iridium began discussions on December 1, 2025, followed by a call between the CEOs on December 15. Further meetings continued in January and February and RocketLab began due diligence in March. After the Amazon acquisition of Globalstar was announced on April 14, the next day RocketLab delivered its first offer of $42.50 in RocketLab shares. An updated offer of $52 per share ($15 in cash and $37 in RocketLab shares) was submitted on June 3 and this was updated further to $26 per share in cash and $26 in RocketLab shares on June 22, before being finalized at $54 (again 50/50 cash and shares) the next day. The dates on which RocketLab began discussions with Iridium are consistent with my assumption that RocketLab was Globalstar’s Party A that made an initial bid but pulled out of the process with Globalstar on December 4.

What is remarkable about the deal is how Iridium was able to push RocketLab’s offer up by over 25% despite there being relatively little competition for Iridium’s assets. The registration statement identifies three parties A, B and C as potential rival bidders, but the only one of these that made a serious offer was Party A, which had initially began meeting with Iridium in November 2025 and conducted due diligence in parallel with RocketLab, before submitting an initial all cash offer of $41 to $45 per share on June 2, which it refused to raise further.

After the Amazon-Globalstar announcement, Iridium’s advisors at Evercore reached out to 10 other potential strategic counterparties of which 4 signed confidentiality agreements but only one (Party B) entered into formal due diligence on May 13 and terminated this two weeks later without making an offer. The only other identified party (Party C), which had not been the subject of Evercore’s outreach, had their CEO reach out to Iridium’s CEO on May 7, but on May 12 said it would not be submitting a proposal to acquire Iridium.

Iridium and Evercore clearly knew at Satellite 2026 about rumors that the Globalstar deal was coming and appear to have cleverly extended the process so that this announcement would inevitably lead to speculation about an Iridium sale and drive up the share price. They also played off the certainty of the cash offer from Party A (without telling RocketLab how low this was) to get a much higher share of cash (“value certainty”) in the RocketLab offer.

So who are Party A, B and C? Party A appears to have been a private equity firm, given the lengthy and detailed due diligence and the emphasis on an all cash offer (including cash settlement of Iridium’s employee equity incentives), as well as the refusal to improve the offer based on Party A’s belief “that Iridium Common Stock was trading at elevated levels amid market speculation about a potential strategic transaction” (i.e. Party A was not a “strategic” bidder).

[EDIT 9/17] We now understand from multiple sources at WSBW that Party A was Amazon, which wanted to buy both Globalstar and Iridium and combine the two companies, although Amazon’s behavior was perplexing, allowing the Iridium negotiations to run on beyond the date on which the Globalstar acquisition was announced, rather than setting a firm deadline, and then seemingly being surprised that this announcement caused Iridium’s share price to rise above the value that Amazon was prepared to pay.

Party B was indicated to “require a strategic partner with which to pursue an acquisition of Iridium” which makes it clear that Party B wasn’t Amazon as some might have assumed. The most plausible “strategic counterparty” in this category is AST SpaceMobile, which would have needed backing from one or more telcos in order to make a viable bid for Iridium.

And Party C is most likely Viasat, which wouldn’t have been in a position to make a bid (and for this reason, as indicated in the registration statement, would probably not have been a “strategic counterparty” that Evercore engaged with), but would have wanted to understand more about what was happening with this key competitor.

So what does this tell us? Firstly, that neither Amazon nor SpaceX were interested in bidding for Iridium. And secondly, that AST isn’t well positioned to make a large cash bid for spectrum (either from Grain or Viasat) without backing from a telco partner. That suggests that the current frantic speculation about a potential bidding war between these companies for Viasat’s spectrum may be misplaced.

[EDIT 9/17] Amazon’s desire to acquire spectrum was greater than we assumed and it appears that Amazon felt this year may have represented the last chance to lock up one of the limited bands of global MSS spectrum that is available. After losing out on Iridium, Amazon might still be interested in securing additional spectrum from Viasat, and it would now not be a total surprise if Viasat received bids from other companies as well, if they take the same view that this is their last best chance to buy more MSS spectrum. Nevertheless, at this point in time, Viasat appears determined to move forward with the Equatys buildout, rather than pursuing a sale process, and has just committed $400M to the venture. Put another way, we suspect that Viasat’s asking price to abandon Equatys would be far greater than any potential spectrum buyer would be prepared to pay.

Viasat’s L-band spectrum is certainly more attractive than Iridium’s more limited amount of spectrum for companies (unlike RocketLab) that are focused on a broadband D2D deployment. But one might conclude that SpaceX and Amazon want to focus on building their satellite networks rather than paying billions for more spectrum in the immediate future. And then they can see whether or not Equatys is able to move forward to deploy its own LEO D2D constellation.

However, many spectrum speculators would now be deeply disappointed if Viasat actually does commit large sums of money to building a LEO D2D network and tries to compete head-on with SpaceX, Amazon and AST. [EDIT 9/17] And they would be even more disappointed if Viasat turns down sizable offers (which could be perhaps $6B-$10B) to buy its L-band spectrum.

Globalstar and Amazon began meeting between March and September 2025 about “a potential strategic collaboration”. Amazon signed an NDA on September 12 and on October 16 submitted an initial price indication of $47.50 per Globalstar share in cash (~$6B). An updated bid of $65 in Amazon stock (or $75 in cash for public shareholders) was submitted on November 12, and then updated to $70 per share in Amazon stock on November 18. Finally on December 1, this was revised to $90 per share in cash or $70 in Amazon stock (later increased to $75), with a maximum of 40% of the consideration in cash, and Globalstar signed an exclusivity agreement on December 6. Although exclusivity expired briefly over the holidays, it was subsequently renewed and extended repeatedly to allow for finalization of the arrangements between Amazon and Apple and the deal was finally announced on April 14.

More intriguingly, the prospectus identifies three other unnamed Parties A, B and C, in addition to Amazon, that made formal or informal bids during the process:

Party A met with Globalstar “from time to time to discuss, among other things, Globalstar’s business and a potential strategic transaction involving Party A and Globalstar” between February and June 2025, and began due diligence on the company in July, making an initial verbal offer in late August and then a formal share-based offer of $4.3B-$5.7B on October 17 (plus some contingent value rights for “future monetization of certain spectrum licenses”). An updated offer of $65 to $75 per share was submitted on November 12 based on equal parts cash and stock (with the stock valued at Party A’s 60 day VWAP), but Party A declined to raise its offer further on December 4, “including because Party A and Customer [i.e. Apple] had not reached mutual agreement with respect to a post-closing commercial arrangement”.

Party B was first contacted by Globalstar’s financial advisers on July 24 and met with Globalstar over the following month to discuss a potential transaction. Party B began due diligence on September 1 and continued through November, submitting a verbal indication of an all-cash deal at $5B-$6B on October 6, followed by a written update on November 14, valuing Globalstar at $9.5B, “less net debt and any obligations owed to Customer (the terms of which were not specified), consisting of $3B in Party B stock, $3B in cash, and the balance in a contingent value right based on the potential sale or monetization of certain terrestrial spectrum assets of Globalstar.” However, Globalstar advised Party B that its CVR-based offer was not competitive and after signing an exclusivity agreement with Amazon, ignored attempts by Party B to reach out again in January 2026.

Party C was first contacted by Globalstar’s financial advisers between August 13 and 20, and on August 21 signed an NDA (which unlike all the other NDAs signed with potential acquirers in that period did not include a “standstill” provision). In mid-September, Globalstar noted that there was “uncertainty regarding Customer and Party C reaching a mutually acceptable post-closing commercial arrangement” and Party C submitted an indication of interest at $4B-$5B on September 23. Party C informed Globalstar’s representatives “that Party C believed that the value of any stock proposed as merger consideration would likely appreciate” and that “Party C was open to pursuing multiple alternative structures, including an acquisition of 100% of the equity of Globalstar, a purchase of spectrum assets or a purchase of Globalstar’s interest in its partnership with Customer”. Party C submitted an updated indicative bid of $8B on October 6, consisting of 50% cash and 50% in Party C stock, with an expectation that all the stock would go to Thermo. However, shortly thereafter, Apple advised Globalstar that “Party C had not engaged in discussions with Customer in recent weeks”.

Then, on October 17, Party C indicated that they were “potentially interested in a transaction that valued Globalstar in the ‘mid teens billions’ (based on Party C’s assertions that the value of any Party C stock proposed as merger consideration would likely appreciate), but that Party C required additional time to discuss a post-closing commercial arrangement with Customer before it could make an updated proposal.” However, by October 29, Party C reported that “Party C and Customer had not progressed discussions with respect to post-closing commercial arrangements between Party C and Customer” and Apple told Globalstar that Apple “thought it was uncertain that a commercial arrangement between Customer and Party C could be reached on terms acceptable to both parties based on Party C’s current proposals.” A November 11 letter from Party C indicated that Party C “remained interested…but required additional time to obtain Customer’s support for a Potential Transaction between Party C and Globalstar and, if Customer was supportive, align with Customer on the terms of a commercial agreement and negotiate the relevant documents.” However, on December 22, Globalstar was told that “Party C was unable to further pursue a Potential Transaction, including because Party C and Customer had not reached mutual agreement with respect to a post-closing commercial arrangement.”

So who are Parties A, B and C? It’s pretty clear that Party C is SpaceX given that the structure they proposed is very similar to that with EchoStar: 50/50 stock/cash, with an expectation that the stock “would likely appreciate”. It is notable that a major barrier to Party C’s bid was the tense negotiations and resulting lack of any commercial agreement with Apple. This confirms that what I posted last September about the pressure SpaceX was putting on Apple was exactly right, rather than “pretty much everything is wrong” as Elon Musk posted in an attempt to head off further reporting on this issue.

Party B appears to be T-Mobile or Deutsche Telekom, which as I’ve noted previously, were involved in the process, after deciding not to pay up for EchoStar’s AWS-4 last summer, but didn’t make a particularly serious attempt to win. The timing also fits with T-Mobile dropping out of the bidding for AWS-4 in July and commencing due diligence at the beginning of September once it became clear that SpaceX was likely to buy EchoStar’s spectrum.

And then Party A appears to be either RocketLab or AST: the giveaway here is the 60 day VWAP between mid September and mid November 2025, which covers a period when the share price of both companies surged to a peak before falling back. In my view it was probably the former, since RocketLab would already have been meeting with Globalstar between February and June 2025 to discuss “other things”, namely the delayed replacement satellites. And RocketLab is less likely than AST to have been interested in retaining “certain spectrum licenses” (presumably terrestrial Band 53 rights).

So what does that imply for the universe of potential spectrum bidders for Viasat’s L-band, as investors have asked me on numerous occasions in recent weeks? There’s little chance that AST and Viasat could reach a meeting of minds over L-band, and RocketLab is off the table after the recent Iridium deal. SpaceX doesn’t need more MSS spectrum right now (and certainly not Viasat’s spectrum which is mostly outside the US): if anything SpaceX will be buying US terrestrial spectrum, most plausibly EchoStar’s CBRS spectrum to facilitate a potential cable partnership for Starlink Mobile V2.

Amongst Parties A, B and C, we are therefore left with T-Mobile/Deutsche Telekom, but after saying on the last earnings call about D2D that “having worked closely with satellite over the last 4 years, we’re just not seeing that incremental differentiation” it appears unlikely that T-Mobile would be interested in investing in satellite spectrum. And does Amazon want to acquire more D2D spectrum right now, especially when it has so much on its plate getting the broadband system into orbit and completing the Globalstar acquisition?

As a result, I think bidders for Viasat’s spectrum are likely to be scarce or non-existent, and despite the challenges that Equatys faces after the loss of RocketLab, it looks like Viasat will continue trying to build a D2D constellation, rather than selling its spectrum to others. Or as Viasat said on the results call last night, “what we’re holding up for is to have the next announcement, which really would be about the Equatys purchase of its initial satellite constellation.”

It has been interesting to compare my numbers with those published by various Wall St analysts, and understand the reasons for these differences. The range of forecasts for the connectivity segment in the reports I’ve seen is pretty wide, but all are a lot higher than my projection of $48B total revs and 46M consumer subs in 2030, with total connectivity revenues and consumer subs as follows:
Deutsche Bank: $72B revs, 82M subs
Moffett Nathanson: $73B revs, 70M subs
RBC: $101B revs, 100M+ subs
Morgan Stanley: $121B, 116M subs.

Unsurprisingly, the most similar forecast to mine is the one from Moffett Nathanson, as Craig Moffett and I share many of the same views about Starlink’s D2D market constraints, and used very similar methods to forecast the consumer market opportunity. The main discrepancy is in the enterprise and government sector, where I’ve broken down the revenue by enterprise, maritime, aviation and government in more detail based on Starlink’s disclosures about pricing and penetration. Deutsche Bank takes a more optimistic view of Starlink’s ability to expand consumer penetration in developing countries via aggressive reductions in ARPU, while also being much more optimistic about enterprise, government and mobility demand. And then RBC and Morgan Stanley take even more optimistic views about demand in all three parts of Starlink’s connectivity business.

One area seemingly missing from most analyst reports is an assessment of how terminal revenues and costs limit the scope for revenue growth (equipment generated $1.5B in revenue last year but is a one-off not a recurring revenue source) and price reductions for end customers (terminals cost an average of over $300 to produce last year). So I end up with higher consumer ARPUs (and slower growth) than many other forecasts, because the need for terminal subsidies (and higher ARPUs to compensate) will constrain demand in developing countries.

The biggest debate is what comes next in mobility. As I’ve noted before, Starlink Mobile’s wholesale ARPUs are far lower than most analysts expect, i.e. tens of cents rather than multiple dollars per customer (of whom less than 10% are active), with partner access fees (effectively minimum commitments) bridging the gap to last year’s reported revenues.

There’s still an opportunity to gain several billion dollars of annual revenue from the Starlink Mobile V2 constellation, if Starlink continues to offer wholesale service to MNOs at a low price (I estimate the effective wholesale capacity price could be under $2 per Gbyte on V2). But that won’t provide a return on $20B of spectrum purchased from EchoStar. And the JV between the US MNOs will try to drive a hard bargain on price by playing off the satellite D2D providers against one another (though AST and Amazon’s offerings are unlikely to match up to what Starlink Mobile V2 will deliver in 2028).

So with MNOs wisely refusing to give Starlink an MVNO deal (thereby retaining their competitive advantage in bundled fixed and mobile services), what is SpaceX going to do? It looks like a small cell plan might be part of the answer: Starlink could copy Charlie Ergen’s plan for Clearwire back in 2013, which involved adding small cells to rooftop satellite (TV) dishes, and speculation is now mounting about perhaps even adding small cells to Cybercabs. I’ve also heard rumors that Samsung might even be building a phone that could carry the Starlink Mobile brand.

One potential source of spectrum for a terrestrial network would be to buy EchoStar’s CBRS spectrum, which is a lot more usable than upper C-band spectrum that won’t be available until 2031 (and isn’t compatible with current phones). And perhaps a deal with one or more cable companies that already use CBRS could be on the table: Starlink broadband could then potentially provide an out of footprint competitive solution that is analogous to FWA for the telcos. However, it remains to be seen how Starlink could address the in-building terrestrial network advantages that AT&T was so keen to highlight “in the stadium…in the hospital…in the high-rise building”.

I pointed out in a recent paper for WIA that D2D complements terrestrial wireless coverage rather than replacing it, and will mostly be just a fallback option for remote and emergency use. So I don’t believe that Starlink can realistically hope to capture a sizeable part of the US wireless market with a D2D-led service, even if this was coupled with small cell coverage, unless it gained access to a terrestrial wireless network with a national footprint. And at the end of the day, while rattling the cage of the telcos may keep analysts and investors on the edge of their seats, SpaceX is all about access to space and the company’s prospects will ride (or die) on Starship.

RocketLab was expected to be an anchor partner in the Equatys joint venture, as I noted back in March, contributing up to $1B of funding in exchange for a contract to build the Equatys satellite buses and launch the satellites. Now RocketLab gets to control its own destiny rather than being under Viasat’s thumb, and instead of competing head-to-head with SpaceX in the D2D market, can focus on what RocketLab and Iridium both perceive to be more promising and protected markets, such as aviation safety and PNT. That’s very clear when the investor presentation emphasizes that Iridium “delivers real time, pole-to-pole global coverage over every ocean, mountain and airway” and builds on Iridium’s recent acquisition of Aireon. Iridium has sold connectivity to the DoD for the last 25 years and is also playing a significant role in support of the Space Force’s PLEO efforts. And if it turns out there is a niche role for a D2D IoT offering, independent of Starlink, then Iridium can support that too.

Now the question is whether Viasat can replace both RocketLab’s bus and launch capabilities, and more importantly its money, to finally get the Equatys venture off the ground. RocketLab’s defection explains why no announcement was made in conjunction with Viasat’s results earlier this month, but this must have been a big shock, given how Space42′s CEO was all but confirming RocketLab’s participation in the joint venture at the Satellite show. It’s hard to imagine that alternative satellite vendors such as MDA would want to simply supply Viasat with buses, or make a substantial financial investment in the venture, or that other startup bus providers would have the money to participate either.

Alternatively, will Viasat have to come to the table with AST and figure out a way to jointly make use of the L-band spectrum? Is that even possible, given that both companies believe their own technical solutions are far superior to the other’s offerings? Does Viasat simply continue what Inmarsat did with LightSquared/Ligado for the last nearly 20 years, and agree to lease more (global) spectrum to AST while taking the view that it will get that spectrum back when AST eventually fails to deliver a viable system? Or will Viasat’s shareholders push for the company to follow Iridium’s path and sell itself to AST? If so they would undoubtedly want many billions of dollars in cash, not just AST shares. It is going to be a long hot summer…

But what is not being talked about is the context of a knife fight behind the scenes between SpaceX and T-Mobile over what T-Mobile will pay for the T-Satellite service after exclusivity expires next month. Because SpaceX was desperate to secure a launch partner and pre-empt Apple’s announcement with Globalstar back in August 2022, the current contract is reportedly only valued at “about $100M” in total since 2022, which means T-Mobile is likely only paying a few million dollars per month for the service. That’s only of order $1-$2 per monthly active user, when less than 10% of T-Mobile customers with T-Satellite included in their plan make use of the service in a given month. Or put another way, about $0.10 per month for each customer with access to the service.

But late last year, when SpaceX looked like it was going to come up short of Elon Musk’s 2025 projection of $15.5B in revenues, it seems the company booked an incremental $632M in (deferred) Starlink Mobile revenues so SpaceX would come close to Musk’s target (SpaceX’s actual reported revenue last year, excluding xAI, was $15.473B). That’s far above the current runrate reported for Starlink Mobile of $85M in 2026Q1.

So now, in order to show growth in 2026, SpaceX needs vastly increased revenues from T-Mobile (5-10 times the current level), when (based on the 2026Q1 runrate) T-Mobile is seemingly paying much less than SpaceX’s other MNO partners. As an example, I understand Rogers in Canada has suggested it might end up paying SpaceX up to CAD100M per year in a market which is about one tenth the size of the US. And we’ve seen numbers being floated like $500M per year for Starlink Mobile to provide service in Iran.

What is clear is that T-Mobile does not believe the service is worth that much, which is why they are being so vocal about how “most of the usage we’re seeing is in national parks” and “we’re seeing a lot less usage than we were originally thinking”, later quantified as “satellite usage is 0.0002% of our total network usage” and “pretty much, no one buys satellite stand-alone”. T-Mobile is even explicitly stating that “we expect that our exclusivity will end”.

T-Mobile is offering some carrots as well as sticks, promoting its new Super Broadband solution with Starlink as a backup for businesses. And of course in Europe, Deutsche Telekom’s support is going to be even more critical after the recent EU 2GHz proposal indicated that Starlink would not be allowed to utilize EchoStar’s spectrum license, even though that license will be extended to spring 2029.

But the proposed D2D joint venture with AT&T and Verizon aims to stop SpaceX from playing off the mobile operators against one another, and the operators have been united in their refusal to consider an MVNO relationship with SpaceX.

This is certainly the right time for the MNOs to hold the line on pricing of the Starlink Mobile service, as it will set an advantageous precedent for future D2D services if the wholesale cost for MNOs is in the low tens of cents per customer per month (an order of magnitude lower than the $2-$3/month assumed by many D2D proponents). And it seems hard to imagine SpaceX turning off the T-Satellite service before its own offering is ready in 2028 (though I wouldn’t completely dismiss the possibility of SpaceX completing the EchoStar transaction early and using some of the Band 70 spectrum on the current Starlink DTC satellites).

Are SpaceX’s current threats enough to change the mobile operators’ minds, even if SpaceX does buy more terrestrial spectrum? It’s always been expected that there will be some direct-to-consumer Starlink Mobile offering, most likely in the form of a smaller version of the Starlink Mini (a Starlink Nano?) operating in MSS frequencies with a much improved battery life, that you can connect to your phone via Bluetooth or WiFi. That’s very different to offering a fully fledged mobile service, when satellite links won’t deliver the data rates and inbuilding penetration expected from a terrestrial network.

But threats to buy a mobile operator or build a terrestrial mobile network seem like a paper tiger, which would do nothing to leverage SpaceX’s “core strengths” as set out in the S-1:

So I look forward to seeing the revenues that SpaceX is able to report from Starlink Mobile in Q2 and Q3, and whether this is really a growth engine for the company in 2026, or if revenues actually fall from 2025 levels. If that’s the case then Starlink Mobile will presumably become another business (like Orbital Data Centers) where the hype is focused on what might be achieved with a next generation constellation in 2028 and beyond, while analysts continue seeking attention by playing up the possibility of SpaceX building, buying or leasing capacity for a fully fledged mobile service.

Rumors have been swirling all week here at the satellite conference in Washington DC and many people have guessed that the winning bidder would be Amazon. That seemed to be the most likely outcome back in January. However, I think that Jay took Amazon’s offer to SpaceX and they probably decided to beat it, in order to further cement Starlink’s dominance of the satellite industry. Amazon already faces severe competitive pressure from Starlink in the broadband market, and a Starlink purchase of Globalstar would block another opportunity for Amazon to broaden its appeal and match Starlink’s D2D offer.

Buying yet more spectrum might not be seen as the wisest course of action for SpaceX, if some international regulators decide that because Starlink controls Globalstar’s MSS spectrum there is no need to grant Starlink additional rights in the 2GHz MSS spectrum acquired from EchoStar. But Brendan Carr’s threats that he will block European satellite operators from the US market if the EU withdraws Starlink’s spectrum rights should carry the day in the near term.

In particular, a two year extension is likely to be granted for the current EU 2GHz spectrum licenses (which is conveniently beyond the November 2028 presidential election). And 2029 is a long way away, given the number of balls Elon Musk has to juggle in the next few years to get Starship flying, meet NASA’s moon ambitions and sort out the challenges at xAI and Tesla.

What is much less clear, is what will happen to Globalstar’s C-3 constellation and the relationship with Apple after a sale. Would Apple continue to pay Starlink hundreds of millions of dollars per year to support connectivity on existing iPhones? Would the C-3 constellation still be completed, especially if it takes MDA another two years or more, by which time Starlink might have its own next generation Starlink Mobile constellation on orbit? That decision could go either way, depending on how confident SpaceX is that Starship will be ready to launch the next gen constellation on time.

Meanwhile, in other news, it seems that Viasat has concluded that it will have to put some of its own money on the line to get started with the 2800 satellite Equatys constellation, with RocketLab likely to be chosen as the satellite bus contractor. Plausibly, RocketLab’s recent $1B fundraising could be used to provide an equity injection into the Equatys joint venture.

And AST is telling people that it will now rely on New Glenn for 9 of its first 12 launches, implicitly confirming the rumors that the company can’t figure out how to stack satellites within the much smaller Falcon 9 fairing. Of course, there’s no way that New Glenn will provide AST with anything like 9 launches this year, so AST’s deployment plans will be pushed out even further.

The level of skepticism about AST at the conference is quite remarkable, but despite Starlink’s dominance, there’s little reason for mobile operators to withdraw their support in the immediate future. Instead, with Starlink’s next generation Starlink Mobile constellation at least two years away and alternatives including AST, Equatys and Lynk all on a similar timetable, MNOs can wait and see, and perhaps pray, that one or more alternatives to Starlink ultimately emerges, and that, in the meantime, more clarity emerges about whether their customers actually care about D2D at all.

That’s perhaps not surprising, because I’m told that an announcement from the EU on the process for reallocating the 2GHz licenses after they expire in spring 2027 remains stalled, and may take several more months to emerge. As a result, the expectation is that the current licenses (held by Viasat and EchoStar) will be extended, probably by two years, to allow time for that process (with appeals and a need for subsequent actions by national regulators) to conclude.

But there is a wider context here to the alignment between Starlink and Deutsche Telekom here, which is seemingly happening with the implicit backing of the German government (and makes me wonder what will happen to Tesla’s factory in Germany this week).

Back in January, news broke that Starlink had struck a fleetwide deal to equip Lufthansa’s aircraft with connectivity. After the prior loss of IAG to Starlink in November 2025, Lufthansa was the last remaining anchor customer for Viasat’s European Aviation Network (EAN).

Lufthansa’s defection fatally undermined the case to retain EAN in its current form and therefore has called into question the need for Viasat to retain 2x15MHz for its 2GHz MSS license. In supporting Viasat’s application for renewal, Lufthansa had even gone as far as to claim that “The EAN is a critical building block in continuing our journey to offer an industry-leading connectivity solution to our passengers.” Of course, it will take some time to replace the EAN terminals on both Lufthansa and IAG, but that just reinforces the rationale for a two-year extension to the current European 2GHz licenses, before any changes take place.

Those changes could potentially reduce both Viasat and EchoStar to paired 10MHz blocks and free up a third license for a European provider (prompting a fight between the AST/Vodafone partnership and the SES/Lynk/Omnispace grouping). But now I wonder if DT might switch its position and suggest cutting Viasat to 2x10MHz, in order to free up 2x5MHz for IoT, while leaving EchoStar/Starlink with the full 2x15MHz. Though whether that would fly with EU regulators is far from clear.

What is interesting is that Lufthansa’s decision to defect to Starlink was apparently very sudden, in fact I’m told that Lufthansa had been negotiating a major deal with another IFC provider for the last couple of years, which was close to being confirmed publicly, and rumors suggest that the German government had a hand in the switch.

Now we have a similar major deal between Deutsche Telekom and Starlink that comes after DT reportedly vetoed T-Mobile US’s plan to buy EchoStar’s spectrum last summer to enhance the partnership between TMUS and Starlink. And last June, DT had been just as unenthusiastic about any changes to the 2GHz band, stating that “Deutsche Telekom AG plans to continue operating the European Aviation Network (EAN) using the MSS 2GHz spectrum beyond 2027…preserving the current spectrum allocation is crucial for the continued operation and economic viability of the EAN.”

So the natural question is “what changed”? I’m told that even after vetoing the TMUS-EchoStar spectrum deal (and then replacing the TMUS CEO while privately characterizing TMUS as “going rogue”), DT continued investigating D2D options and was one of three companies that looked at Globalstar when that asset was put up for sale last fall (the other two being SpaceX and Apple).

But nothing happened there, and now DT has decided instead to strike a major partnership with Starlink, preferring to rely on the 2GHz MSS band over Globalstar’s Big LEO spectrum. And there’s even a DT panelist (Jaroslav Holis) scheduled to speak at the Equatys event on Wednesday which raises the question of what DT might have been exploring there. So I’m left wondering whether there are wider German political factors behind the decisions of both Lufthansa and Deutsche Telekom to reverse themselves in short order.

With Deutsche Telekom apparently getting cold feet about buying spectrum for D2D, and Verizon not yet at the table, that meant the most likely scenario was for EchoStar to continue moving forward with its own constellation, in order to keep control of the whole AWS-4 block and significantly constrain Starlink’s D2D capacity in the US (while having the opportunity to monetize the spectrum in urban areas through leases to a wireless operator like Verizon).

But ROI has never been the primary determinant of SpaceX’s decisions, when the opportunity presents itself to dominate an industry and force competitors out. That’s why we are seeing aggressive actions from Starlink in the satellite broadband market, lowering prices for hardware and service in both the consumer and professional markets to make Amazon Kuiper’s entry harder (including a new unlimited maritime plan for merchant vessels at only $2500 per month, which will also undermine Viasat’s NexusWave).

And in this case, by spending $17B, SpaceX has not only persuaded EchoStar to give up its D2D plans but has now made it much harder for any competitor to move forward when they can’t possibly compete with SpaceX’s speed in bringing new satellites to market. That was evident in the article published by The Information in May, where Apple staff working on the D2D project with Globalstar expressed concerns that their bosses would cancel the effort and decide to partner with SpaceX instead. And we’ve seen more on that front in recent months, as Globalstar’s new satellites have been delayed, and Apple was apparently forced to support Starlink on the iPhone 13 in order to secure a new launch slot.

It shouldn’t be ignored that just like in fall 2022, the SpaceX announcement comes right before Apple’s own event tomorrow to announce its new iPhone. So while this might not be on the agenda tomorrow, decisions about the future of the Apple-Globalstar partnership and the new C-3 constellation will be on everyone’s minds. The cancellation of the EchoStar D2D constellation was already a major blow for MDA, but any decision by Apple to pull back from the C-3 constellation would be even more devastating.

SpaceX especially wants Apple to cooperate instead of pursuing the C-3 constellation because the H-block and AWS-4 spectrum, that SpaceX is now acquiring from EchoStar, is not supported by any current phones (EchoStar’s Band 66 and Band 70 used different frequency pairings). Thus support from device manufacturers will be needed to get the new capabilities enabled by this spectrum into consumers’ hands in the near term. Of course if Apple doesn’t come around, then there’s always the possibility that SpaceX will announce a “Starlink phone” as Apple executives worried about in the May article.

In recent years, Musk has also plotted the ultimate challenge to Apple, said a person with direct knowledge of his thinking: building his own phone to get around Apple’s gatekeeper position in the market. Musk has discussed Tesla building the phone and providing satellite connectivity through Starlink, the person said.

Musk hasn’t kept his openness to making a smartphone secret. He has publicly toyed with the idea on social media at times, but he has also made it clear he doesn’t want to deal with the headaches of such a monumental effort.

“The idea of making a phone makes me want to die,” Musk said at a Trump rally in Philadelphia last October. “If we have to make a phone, we will. But we will aspire not to make a phone.”

And as far as other competitors go, AST is already struggling with enormous delays, which are now even worse than the company indicated in mid August, after the FM1 satellite wasn’t ready to ship at the end of August as promised during AST’s Q2 results. And AST needs to raise over $400M in the next few weeks to make the $420M payment due to Viasat at the end of October. The one good piece of news for AST from this deal is that it very likely means EchoStar won’t retain its EU 2GHz license (though there will undoubtedly be litigation if it is cancelled), leaving AST/Vodafone in competition with SES/Lynk for what will presumably by a paired 10MHz license (assuming Viasat retains its own paired 15MHz license).

It’s also unclear what Viasat will do next, as the company hoped to secure financial backing from UAE-based Space42 to build its own LEO L-band network. While I don’t think a formal deal was likely to be announced next week in Paris, this announcement probably gives Space42 further pause about whether it makes sense to challenge Starlink in the D2D market, especially as the expectation was for Space42 and the UAE government to put up most of the funding.

Finally, I think we can now look to EchoStar to gradually wind down the rest of its operations and sell off its remaining spectrum. The remaining major block is AWS-3, which Verizon might pick up in the next few months, potentially at a discount to the $10B EchoStar paid, especially if Verizon takes on the AWS-3 reauction obligations. And then it would be reasonable to assume that DISH DBS would merge with DirecTV and Hughes could eventually be sold (perhaps to a private equity buyer?).

The big prize now is EchoStar’s collection of midband assets in the AWS-3, H-block and AWS-4 bands, which could collectively be valued at as much as $30B. Semafor suggested that a three-way deal between AT&T, T-Mobile and EchoStar had been discussed under which AT&T and T-Mobile “would have swapped some of their own spectrum holdings”, but later indicated that “T-Mobile’s ultimate owners, Deutsche Telekom, tapped the brakes”.

This has caused speculation to focus on Starlink and even Kuiper as potential buyers of these assets, but what many articles are getting wrong is the suggestion that this is because (as Semafor put it) Starlink “wants its own network to provide cell coverage, something that would disrupt the stranglehold that AT&T, Verizon, and T-Mobile have on the US market”.

That’s a complete misunderstanding of the Direct-to-Device (D2D) business, which (despite the nonsense promulgated by some AST SpaceMobile investors) is limited to much slower speeds and far less capacity than terrestrial networks. It’s a simple matter of physics that communicating from your smartphone to a satellite hundreds of miles up in space will be less efficient than communicating with a cell tower a mile or two away and that means D2D is not a true substitute for terrestrial cellular service.

The consequence of this lower throughput and capacity is that D2D can’t generate the same revenue from each MHz of spectrum in space as a terrestrial operator on the ground, and so D2D operators can’t afford to pay as much to acquire spectrum. That’s why we’ve seen increased interest in cheaper MSS spectrum, both from Apple investing in Globalstar and more recently AST SpaceMobile bidding for Ligado’s spectrum.

But EchoStar’s mooted $30B price tag is only achievable by buying this spectrum for use in a terrestrial network, which is why Starlink has been trying to persuade the FCC to award it some of EchoStar’s spectrum for free. If that doesn’t work out then Starlink needs T-Mobile to pay the vast majority (if not all) of the $30B that EchoStar is demanding. So if T-Mobile steps back and we see FCC Chairman Carr accepting EchoStar’s offer to sell spectrum (and canceling the idea of a 2GHz MSS NPRM that might open up the band for sharing with Starlink), there’s no realistic prospect of Starlink and EchoStar agreeing on price.

We’d guess that Deutsche Telekom might want to wait for more evidence of the success or otherwise of T-Mobile’s D2D collaboration with Starlink before paying tens of billions for spectrum that they don’t really need, mainly so Starlink can improve the capacity of its D2D network. But if T-Mobile did in the end decide to bid, then either Starlink could buy the H-block (which cost EchoStar only $1.5B) and extend its existing G-block SCS network from 5x5MHz to 10x10MHz, or T-Mobile could offer Starlink access to some of the AWS-4 spectrum in rural areas for D2D.

However, there’s also an alternative path for T-Mobile and AT&T to just swap the 600MHz holdings that AT&T has now agreed to buy from EchoStar, for T-Mobile’s C-band spectrum assets, and not do any further deal with EchoStar.

If T-Mobile did buy all of EchoStar’s midband spectrum, then of course EchoStar’s planned D2D constellation would be abandoned. But there’s no reason to treat that as the default outcome. If instead Verizon puts in a bid for EchoStar’s midband holdings, then it isn’t allied with Starlink and wouldn’t want to risk the possibility that the FCC grants Starlink access to the 2GHz MSS band for D2D and impairs Verizon’s terrestrial usage plans.

So the best way forward would be for EchoStar to go ahead with its own proposed D2D constellation in order to keep exclusive access to the 2GHz MSS band in the US. Then Verizon could buy EchoStar’s AWS-3 and H-block holdings and lease AWS-4 from EchoStar in urban areas, while EchoStar coordinates D2D usage in rural and remote areas outside the reach of Verizon’s towers.

And finally if neither T-Mobile nor Verizon show up with an acceptable bid, then EchoStar will still want to preserve its MSS spectrum rights (and the associated terrestrial spectrum value in the US) by going ahead with the planned D2D constellation. Thus there are four possible scenarios and only in the first of them would EchoStar’s D2D constellation be abandoned:

1) T-Mobile buys all of EchoStar’s midband spectrum (and shares some with Starlink)
2) T-Mobile just does a swap with AT&T (600MHz for C-band)
3) Verizon buys EchoStar’s AWS-3 spectrum and leases AWS-4 in urban areas
4) No one shows up with $30B to meet EchoStar’s asking price.

On balance, assuming FCC Chairman Carr accepts the current EchoStar-AT&T deal, it therefore seems more likely than not that at least the first stage of EchoStar’s constellation will be built. And analysts who assume it won’t be and that Charlie Ergen is simply planning to sell up and retire might instead find themselves watching this show for many more years to come.