Why the operators are doing this

The three companies in the table are three different kinds of business

Every figure in Exhibit 1 is from an SEC filing, tagged and dated; none is from an investor presentation. The three companies compared are satellite operators whose filings cover their satellite business on its own. SpaceX has filed since June 2026 but reports its phone service inside a larger segment, which is covered below.

The three columns are not three stages of one company. Iridium is a profitable incumbent, with $871.7M of revenue in FY2025, the only positive net income in the table and a mature constellation maintained on $100.3M a year of capital expenditure. Globalstar is a wholesale business of a different kind, with $273.0M of revenue, a net loss of $8.7M, research spending of $6.1M and spectrum it owns rather than rents. AST SpaceMobile is a construction project. It spent $1,064.7M in FY2025 against $70.9M of revenue, funded from cash of $2,335.7M, which works out at a little over two years of spending at that rate.

Cash from operations makes the same point. Iridium generated $400M from operations in FY2025 and Globalstar $622M, so both pay their own way. AST ran an outflow of $72M in the year and $145M in the first half of 2026. Its cash fell from $2,335.7M to $2,288M over that half, and it carried $2,963M of long-term debt at 30 June 2026. Two of these businesses fund themselves; the third draws on money it has raised.

AST’s capital spending rose sixfold in a year

Capital expenditure is the fastest-moving line in AST’s accounts: $174.1M in FY2024, six times that at $1,064.7M in FY2025, and $859.2M in the first half of 2026 (Exhibit 2). Revenue for the same periods was $4.4M, $70.9M and $46.0M. The company is in the middle of a build, and the build is the strategy rather than a preliminary to it.

The rest of AST’s column follows from that. A net loss of $341.9M is what construction on this scale produces. The $28.1M of research spending looks small only beside the hardware spend. The $2,335.7M of cash matters chiefly as a measure of how long the build can continue at the current rate; margin is a question for the day construction stops.

The three companies therefore deserve different questions. For Iridium and Globalstar, which are being run, the questions are margin, retention and renewal. For AST, which is being built, they are how much is left to build and who pays for it.

SpaceX’s filings do not separate its phone business

The largest constellation in the market is missing from Exhibit 1 even though SpaceX is now a public filer. It filed a Form S-1 on 20 May 2026, a prospectus on 12 June 2026 and its first Form 10-Q, for the quarter to 30 June 2026, on 4 August 2026. The 10-Q details the Connectivity segment, which had $4,291M of revenue, $1,656M of income from operations and $1,367M of capital expenditure in the quarter, and shows $93,522M of cash and cash equivalents at 30 June for the group as a whole. Starlink Mobile revenue, however, is reported inside the segment’s Enterprise & Government line of $1,806M, alongside aviation, maritime and government work, and nothing in the filing separates it out (Exhibit 3).

We have not filled the blank with an estimate. It would be the only unfiled number in an argument built on filings, and it would be the number everyone quoted. In our view a market model that treats all four operators as equally knowable is wrong at the foundation; a sound one has three measured companies and one blank, and says so.

Renting is the only route the capital expenditure line leaves open

Part 1 showed that AST and SpaceX hold leases rather than licences, so neither owns the spectrum its consumer service depends on. On its own that looks like a weakness; against the capital expenditure line it looks like the only route open. We see three reasons the arrangement holds, all of them financial.

• Crowding out. A company spending a billion dollars a year on hardware cannot also buy the band it flies on. Buying AST’s 24 MHz nationwide would cost roughly $5.4B to $7.2B at the prices in part 4, which is two to three times its cash. A lease keeps the capital in the constellation, where the business needs it.
• Optionality. A lease can be exited and a licence purchase cannot. For a service that has not yet proved consumer demand, reversibility has value: renting turns an irreversible bet on a market into a recurring cost that can be stopped.
• The carrier’s side. The carrier lends spectrum it already owns, gains coverage it could not build economically, and keeps billing its own subscriber. The satellite operator gains reach without winning customers one at a time. Each side supplies the part of the business it is better placed to supply, which makes the arrangements wholesale rather than retail (Exhibit 4).

The arrangement has a built-in constraint: the satellite operator’s revenue depends on a partner who owns both the spectrum and the customer. Globalstar shows the same logic from the other end. It owns its spectrum outright and still earns most of its revenue wholesaling capacity to one large consumer partner rather than selling retail, so its accounts show substantial revenue and almost no margin. Owning the spectrum removes the landlord; it does nothing about who owns the customer. Both incumbents have since agreed to be bought. Amazon’s agreement of 13 April 2026 pays Globalstar holders $90.00 a share in cash or 0.3210 Amazon shares. Rocket Lab’s agreement of 28 June 2026 pays Iridium holders $27.00 in cash plus Rocket Lab stock at a ratio of 0.2400 to 0.4000. Part 6 sets out the terms and who is paying.

The tenancy comes up for renewal on one day in May 2028

The dates are exact. All 1,197 of AST’s lease authorisations were granted on 21 April 2026, and every one expires on 9 May 2028 (Exhibit 5). A footprint assembled in a single administrative act comes up for renewal in a single administrative act, so the whole American business rests on one calendar entry.

Whatever is built on that footprint has about two years to prove itself before the commercial terms are renegotiated. The counterparties at that table are the carriers whose customers the service serves.

The other tenant is buying its way out. SpaceX expects to own 65 MHz from late 2027, and on 8 October 2026 it agreed to buy a second block, up to 14 MHz of paired 800 MHz spectrum from Grain Management, at a price neither party has stated. The capital constraint that keeps AST renting does not bind SpaceX, which held $93.5 billion of cash on 30 June 2026. Part 4 sets out both purchases and part 6 the balance sheets behind them.

Anyone modelling this business should treat the date as an input rather than a footnote. It bounds the window in which evidence on usage, pricing and retention has to be produced, and anything meant to improve the renewal terms has to exist before it. The alternative is to stop being a tenant, which has a price of its own; part 4 works out what owning the spectrum would cost.

Terms used

• Capital expenditure (capex). Cash paid for property and equipment in the period, as each company tags it in its filings. It is not the same as constellation cost, and the tags differ between filers.
• XBRL company facts. The machine-readable tags behind every SEC filing, published at data.sec.gov. Reading them gives the filed number rather than a summary of it.
• Wholesale. Selling capacity to another company, which sells the service to its own customers under its own brand and bill. Retail sells to the end customer directly.
• Lease authorisation. The FCC record permitting one company to use another’s licensed spectrum for a stated term. AST’s 1,197 are its entire American footprint.
• Runway. Months of cash at the current rate of spending before new financing is needed. We calculate it from the filed half-year figures rather than from guidance.