The Constellation Is Fine. The Income Statement Isn't.

A special report on AST SpaceMobile at $66.

Position disclosure: no position in ASTS, long or short, now or ever. Our April 20 piece said watching, no position. Still true. Nobody paid for this report and nobody is positioned around it. That is the point of it.

This report is free, ungated, and built to be forwarded. Quote anything with attribution to Shawarma Capital, research.shawarmacapital.net. All figures marked to the July 15, 2026 close of $66.31 unless dated otherwise.

Key findings, each one verifiable in a filing:

• Satellite service revenue in Q1 2026 was $1.3 million. The market cap is $20 billion. A pre-revenue multiple is not the indictment, every infrastructure story prices the future. The finding is the race underneath it: the dilution is arriving on schedule and the revenue is not.

• 91 percent of the quarter's revenue was hardware, and more than half of the total was gateway equipment billed to a related party.

• In August 2025 the CEO confirmed a fully-funded plan. Within a year the company priced $2.1 billion in convertible notes across two offerings.

• The company targets approximately 45 satellites in orbit during 2026. It has nine. Of its three named launch providers, Blue Origin blew up its own pad in May and is grounded until December, ULA has yet to fly an AST mission, and the only one actually flying the manifest, SpaceX, competes with AST for the same customers.

• The July convert converts at $79.57. Convert arbitrage hedging is a standing short position above the tape. The stock does not escape the strike zone until the income statement does.

Executive summary, one page

AST SpaceMobile is building a constellation purpose-built to deliver true broadband, not just texting, to unmodified smartphones. Starlink's direct-to-cell reaches the same phones, but at messaging-grade speeds today, while AST's far larger arrays target actual broadband and it holds the first US commercial authorization for it. Over the last ninety days it launched three satellites successfully, won that FCC authorization for commercial US service, signed a $926 million sovereign-network deal in Japan, and grew its partner roster to roughly 60 mobile operators covering more than 3 billion subscribers. The technology case got stronger. The stock still round-tripped from $85 through $133 back to $66, and the reason is arithmetic, not sentiment.

Three exhibits carry the report.

Exhibit A

Exhibit A. The $14.7 million question. First-quarter revenue printed $14.7 million against the $24 to 26 million the ramp implied. Inside that number: $13.4 million was hardware, gateway equipment sold to partners, and $7.9 million of that was billed to a related party. Satellite connectivity service, the thing the $20 billion market cap is priced on, produced $1.3 million. Nine percent of a $14.7 million quarter. The 2026 guide of $150 to 200 million is only about half covered by contracted backlog, and the company now points to 2027 for commercial service at scale.

Exhibit B

Exhibit B. The manifest math. Management targets approximately 45 satellites in orbit during 2026. Today there are nine functioning spacecraft, five Block 1 and four Block 2, after BlueBird 7 was lost in April. Three more are slated for early August. That leaves roughly 33 satellites to fly in under five months, at a cadence the company has never demonstrated, while its launch options narrow. AST names three providers, SpaceX, Blue Origin, and ULA, but Blue Origin's New Glenn destroyed its own launch complex in a May anomaly and does not return to flight before December, and ULA has yet to fly an AST mission. Every near-term launch that actually flies now routes through SpaceX, a company whose own Starlink direct-to-cell product competes with AST for the same subscribers.

Exhibit C

Exhibit C. The dilution ratchet. In August 2025 the CEO told investors, quote, we are confirming our fully-funded plan to deploy 45 to 60 satellites into orbit by 2026. Since that sentence: a $1.075 billion convertible in February 2026 and a $1.0 billion convertible priced July 15, 2026, at 1.625 percent due 2034, converting at $79.57, with $96.9 million of the new money spent on a capped call to push effective dilution to $149.20. That is $2.1 billion of fresh converts inside a year of the words fully funded. Pro forma debt now stands near $4 billion against roughly $4.4 billion of cash, the share count is 299 million and rising, short interest is 21.7 percent of float and at a nine-month high, the stock just left the Russell 2500, and the company president filed sales at $126.64 within a day of the all-time high. The quarter's net loss was $191 million. The capital machine is working exactly as designed. That is the problem: it is designed to run on your shares.

Verdict. The thesis is alive and the price of admission finally reflects the risk. We grade our own April call at the bottom, publish the updated scenario table, and list the three tripwires that would make us buyers. Until service revenue exceeds hardware revenue in a printed quarter, this is a launch-cadence option, not an infrastructure stock.

I. Grading our own April work first

The April 20 piece made five calls. The record, honestly scored.

Right: the satellite loss was noise. We wrote that BlueBird 7 was a launch vehicle failure, not a technology failure, that insurance would cover it, and that the next launch was with SpaceX. All confirmed. BlueBirds 8, 9, and 10 went up clean on a Falcon 9 on June 17, the Block 2 design with 2,400 square foot arrays flying as specified.

Right: the Rakuten overhang resolved bullishly. April's report treated Rakuten's $270 million share sale as portfolio mechanics, not thesis abandonment, and said watch the commercial agreement instead. On July 2 Rakuten committed $926 million alongside AST for a sovereign direct-to-device network in Japan, with government approval following a day later. The seller of April is the anchor tenant of July.

Right, with an asterisk: the technology. The company demonstrated 98.9 megabits per second to an unmodified smartphone, the best direct-to-device figure ever published, and won the FCC's first Supplemental Coverage from Space authorization for commercial service, covering up to 248 satellites. The moat claims are converting into regulatory fact.

Wrong: the revenue trajectory. We relayed an implied Q1 of $24 to 26 million from the Q4 exit rate. The actual print was $14.7 million, and the composition was worse than the size, as Exhibit A details. The Q4 2025 number that anchored the ramp story was itself heavy with gateway hardware and milestone recognition, not recurring service. We passed along the annualized-run-rate framing without decomposing it. That was the error: right numbers, wrong reading.

Wrong: the base-case timing. April's scenario table put base-case 2028 revenue at $800 million on constellation scale arriving through 2026 and 2027. The company has since told investors commercial service at scale is a 2027 event, and the street's Q2 estimate sits at $25 to 40 million. The revenue curve shifted right roughly a year. The scenario table at the bottom reprices accordingly.

Ninety days is a short grading window. It was long enough for the market to reprice the same facts from $133.09, the May 28 all-time high, to $65.62 at the low. What follows is the audit of what actually changed.

II. Exhibit A in full: what the revenue actually is

The Q1 2026 filing, reported May 11, decomposes as follows.

Read that middle line again. More than half the quarter's revenue was gateway hardware billed to a related party. This is not an accusation of anything improper, related-party gateway sales to MNO partners who are also shareholders are a documented part of the model, and gateway deployments are a necessary precondition for service revenue in each market. But an investor pricing a $20 billion company on a revenue line should know that the line is currently dominated by one-time equipment deliveries to affiliated buyers, not by satellites billing subscribers.

The service line, $1.3 million, is the entire current commercial reality of space-based cellular broadband at AST. It will grow, and to be clear about the frame, a $20 billion valuation on nascent revenue is not itself the finding. Amazon traded at absurd trailing multiples for a decade and the future showed up. The finding is the race: the future revenue is now dated 2027 by the company's own offering documents, while the dilution funding it is dated February 2026, July 2026, and, if the pattern holds, whenever the next launch contract needs signing. When a stock prices the future, the only question that matters is which future arrives first. Right now the converts are winning.

The full-year guide of $150 to 200 million was reaffirmed on May 11, with management stating approximately 50 percent is covered by existing contracted backlog. Half of a guide is a plan; the other half is a hope, and the mix inside the contracted half skews toward more gateway equipment and government milestones, not consumer service. Meanwhile the July offering documents moved commercial service at scale into 2027, a slip the market punished within hours, and the street's Q2 consensus of roughly $35 million at the midpoint implies the back half must produce over $100 million to reach even the guide's floor.

Operating costs make the clock audible. Total Q1 operating expenses were $164.1 million, $91.2 million after stripping stock compensation and depreciation. Net loss attributable to common was $191.0 million, one quarter. Gross property and equipment stands at $1.8 billion. Every quarter of 2027-timeline slippage is roughly $200 million of GAAP loss and a proportionate draw against the cash pile, and the cash pile is borrowed.

III. Exhibit B in full: the manifest against the calendar

The deployment record, from the company's own announcements.

From 12 planned in August to the stated 45 requires roughly 33 additional satellites in under five months. At three per Falcon 9, that is eleven dedicated launches, better than one every two weeks, from a standing start of one launch per month at best. The company's Midland and Texas facilities are producing at a claimed rate of phased arrays for more than ten satellites per month, and BlueBirds 11 through 33 are stated to be in various stages of assembly. Production may genuinely be there. Orbit is not production. Orbit is launch slots.

And the launch market just got smaller. Blue Origin's New Glenn, the vehicle that lost BlueBird 7 in April, suffered what its own filings call a devastating anomaly in May that destroyed the launch pad. NASA is assisting the investigation, an engine issue is suspected, and the company has guided return to flight to December 2026 while it scraps and rebuilds the complex. Every near-term AST launch therefore routes through SpaceX.

Sit with that dependency. SpaceX operates Starlink direct-to-cell with T-Mobile, the direct competitor for AST's core product. AST's manifest, its cadence, and therefore its 2026 target and its 2027 service date now run through the pricing desk of its principal rival, with Blue Origin grounded until year-end and ULA yet to fly a mission for it. The July offering language about using proceeds to secure additional access to orbit, and the analyst chatter about buying a rocket company outright, are the company saying the same thing with a billion dollars: launch is the binding constraint, and it is currently rented from the competition.

For completeness, the FCC authorization covers up to 248 satellites, and management holds that roughly 45 to 60 are needed for continuous US coverage. At the demonstrated 2026 cadence, nine satellites in six months including a loss, continuous coverage is a 2028 story without a step-change in launch procurement. That is what the $1 billion is for. Which brings us to what it costs you.

IV. Exhibit C in full: the ratchet

The capital stack, before and after the last ninety days.

Start with the sentence that anchors the exhibit. August 2025, the CEO, on the record: we are confirming our fully-funded plan to deploy 45 to 60 satellites into orbit by 2026 to support continuous service in the US, Europe, Japan, and other strategic markets. Fully funded. Since then the company has priced two converts, $1.075 billion in February 2026 and $1.0 billion on July 15, 2026, roughly $2.1 billion of new paper inside a year of that quote, while the 45-satellite half of the same sentence slipped as Exhibit B shows. Neither raise is scandalous on its own. Together with the quote they are a pattern, and the pattern is that funding statements from this company have a shelf life measured in quarters. Price the next one in.

The July 15 convert: $1.0 billion principal, 1.625 percent coupon, due February 2034, conversion at $79.57, a 20 percent premium to the $66.31 reference. The company spent $96.9 million of the proceeds on a capped call raising the effective dilution threshold to $149.20. An additional $150 million greenshoe is available. Stated use of proceeds: growth initiatives and securing additional access to orbit, including potential partnerships and acquisitions. The stock fell roughly 12 percent around the announcement and pricing.

The mechanics matter as much as the amount. Convertible buyers hedge by shorting the common, which is part of why short interest sits at a nine-month high, and that hedging flow acts as a standing offer above the tape. Two billion dollars of converts struck at $79.57 and, from the February paper, higher, means systematic selling pressure into every rally between here and the strikes until the business generates cash. The retail complaint that the offering caps the stock is not paranoia, it is the arithmetic of how convert arbitrage trades. The cap lifts when the income statement lifts it, not before.

This is the second billion-dollar-class convert in a year, and it will not be the last, because the model consumes capital by design: satellites are capitalized, launches are purchased, gateways are deployed ahead of revenue, and the income statement loses $191 million a quarter while service revenue is $1.3 million. Management calls the balance sheet a fortress. It is, and the fortress is mortgaged: roughly $4 billion of debt against a business whose recurring revenue would not cover one quarter's interest, secured by the promise of 2027.

The supporting cast of signals, each small, all pointing the same way. President Scott Wisniewski filed sales of 25,904 shares at $126.64 on May 27, within a day of the $133.09 top, roughly $3.3 million. CEO Abel Avellan's May filing was a tax-withholding disposition at $113.41, not an open-market sale, and we flag the difference deliberately. The stock exited the Russell 2500 in the July reconstitution, removing passive bid. Short interest sits at a nine-month high with 2.6 days to cover, which cuts both ways: it is fuel for violent rallies on good launches and a standing vote that the arithmetic above is widely understood.

One more structural note. Amazon closed its $11.6 billion acquisition of Globalstar in mid-July, buying Apple's satellite partner outright, days after the FCC signaled an August vote on opening more direct-to-device spectrum. SpaceX flies its own constellation, Amazon now owns one, and both have consumer distribution AST will never match. AST's counter is physics, bigger apertures delivering broadband speeds the others cannot, and carrier alignment, roughly 60 MNOs plus the proposed AT&T, T-Mobile, Verizon joint venture that pools spectrum for exactly this service, with AST positioned as a preferred capacity provider. The competitive story is genuinely two-sided now. It was one-sided in April.

V. What got stronger, stated as plainly as the holes

An audit that only lists holes is a short report, and this is not one. Since April 20:

• Three satellites launched flawlessly on June 17, the largest commercial phased arrays ever flown, designed for roughly 200 megabit peak speeds, nearly double Block 1.

• The FCC granted the first commercial SCS authorization in US history, up to 248 satellites. The regulatory moat is real and it is AST's, not Starlink's.

• Rakuten converted from overhang to anchor: $926 million committed for a sovereign Japanese network, government approval in hand, on top of Japan's earlier equity history.

• The MNO roster reached roughly 60 operators spanning 3 billion subscribers, adding Telus and Axian, with ground integrations under way in 17 countries covering 2.9 billion people.

• Three new US government awards since March, on top of the existing defense relationships, and a $100 million capital advance securing long-term L-band spectrum access through Ligado.

• The 98.9 megabit demonstration to an ordinary smartphone stands as the industry record.

• Vodafone's European joint venture named Spain for its rollout with 2027 service, New Zealand approved a gateway, and European emergency-service demonstrations completed ahead of the August launches.

The asset being built is unique, the demand side keeps signing, and the regulator keeps saying yes. Nothing in this report disputes the destination. Everything in this report is about the fare.

VI. The repriced scenarios

April's table assumed 2026 constellation scale and $800 million of 2028 base-case revenue. Both slipped roughly a year. The update, our model, our probabilities.

Probability-weighted value: roughly $89. Against a $66 tape that is 35 percent of expected upside, materially less asymmetric than the same math looked in April at $85 against a $127 weighted value, because the bear case is fatter and nearer. The option is cheaper and the odds are worse. Both things happened.

What would make us buyers, three tripwires, any one of which converts this from option to position:

1. Service revenue exceeds product revenue in any printed quarter. The business inverts from hardware-and-hope to infrastructure.

2. A sustained launch cadence of six or more satellites per quarter for two consecutive quarters, from any provider mix, proving orbit access is procured, not promised.

3. A definitive capacity contract with the carrier joint venture or a named MNO with committed dollar minimums disclosed, converting the 60-partner roster from memoranda into backlog.

And the standing kill signal, published in advance: a third convertible or equity raise before commercial service revenue passes $50 million in a quarter would confirm the ratchet is the business model, and we would say so in print and stop covering the name as an investment candidate. Management has already spent the benefit of the doubt once, at fully funded. There is no second one.

VII. Verdict

In April we wrote that the market was pricing a thesis break and the math said otherwise. The market then agreed with us for five violent weeks, doubled the stock, and spent the next seven taking it all back as the filings caught up. Both moves were the same mistake in opposite directions: pricing the story instead of the composition.

The composition today: $1.3 million of quarterly service revenue, nine satellites, SpaceX as the only reliable launch path and also the chief competitor, $4 billion of debt, a 2027 service date, and the best direct-to-device broadband technology and regulatory position in the sector. That is a real company building a real thing on rented rockets and shareholder patience.

At $66 you are no longer paying the euphoria tax. You are paying the dilution tax, on a schedule management just published at $79.57 a share. We will pay neither until one of the three tripwires fires. The constellation is fine. The income statement isn't. Watching, no position, and the next hard look is the August 10 print and the BlueBird 11-13 launch in the same week.

Not investment advice. Do your own work.

Methodology note

Every figure in this report traces to a primary source: SEC filings, company press releases, FCC records, and Form 4 filings, with market data from the July 15, 2026 close. Scenario outputs are Shawarma Capital model estimates, not company guidance. We held no position, long or short, at any time during the preparation of this report.

References

1. AST SpaceMobile Q1 2026 business update and results, May 11, 2026. Business Wire and SEC filings. Revenue composition, opex, cash, guide, FCC SCS authorization, MNO roster, 45-satellite target.
   1a. AST SpaceMobile Q2 2025 business update, August 2025, SEC Form 8-K exhibit. CEO quote, confirming our fully-funded plan to deploy 45 to 60 satellites into orbit by 2026. February 2026 convertible notes offering, $1.075 billion, SEC filings and Investing.com coverage.

2. AST SpaceMobile 10-Q, quarter ended March 31, 2026. Products and services revenue split, related-party disclosure.

3. Convertible notes pricing release, July 15, 2026. Business Wire. Terms, capped call, use of proceeds.

4. BlueBird 8-10 launch release, June 17, 2026, and BlueBird 11-13 scheduling release, June 23, 2026. Business Wire.

5. Rakuten-AST Japan sovereign network coverage, July 2-3, 2026. Yahoo Finance and company statements.

6. Blue Origin New Glenn May anomaly and pad rebuild coverage, July 7-14, 2026. MSN, autoevolution, Yahoo Finance. Return to flight December 2026.

7. Form 4 filings, AST SpaceMobile, May-July 2026. SEC EDGAR. Wisniewski sale May 27; Avellan withholding disposition May 30.

8. Amazon-Globalstar acquisition coverage, July 10-16, 2026. Stocktwits, Il Sole 24 Ore syndication. $11.6 billion, Apple stake included.

9. AT&T, T-Mobile, Verizon direct-to-device joint venture coverage, May 14-20, 2026. Reuters, Light Reading, Via Satellite.

10. Service timeline to 2027 and acquisition language, July 15-16, 2026. PCMag, Aviation Week, Yahoo Finance.

11. Russell 2500 deletion, July 10-13, 2026. simplywall.st, Yahoo Finance.

12. Short interest and options data, July 14, 2026. Stocktwits, MarketBeat. Market data from Yahoo Finance, July 15, 2026 close.

13. Prior coverage: One Bird Down, Constellation Intact, Shawarma Capital, April 20, 2026, at research.shawarmacapital.net.