AST SpaceMobile

IThe business

AST SpaceMobile is building a mobile network in space. Its satellites, called BlueBirds, carry very large flat antennas, and the aim is to reach an ordinary phone with no new hardware, so that a person outside the range of towers keeps a signal. AST does not serve the public directly. Its customers are mobile carriers, which its filings call mobile network operators, and it plans to share the revenue with them.⁴³

The newest satellites, called Block 2, have an antenna of about 2,400 square feet. That is three times the size of the Block 1 antenna and, by the company’s account, about the size of a tennis court. They are designed for up to ten times the bandwidth of Block 1.⁴⁵ AST has designed its own chip for them, called the AST5000, and says it has about 95% of the Block 2 subsystems under its own design or control. The chip goes into satellites from BlueBird 34. The target is six satellites a month from two Texas factories in December 2026.³²⁵ The satellite goes up folded, then unfolds in orbit. Seven Block 2 satellites have unfolded, and one, BlueBird 7, was lost on 19 April 2026 when a Blue Origin rocket left it in too low an orbit.⁴¹¹

What it earns today

None of today’s revenue comes from the service. The 10-Q says so in plain words: the service has not launched and has produced no revenue.⁴ What AST earns now is a bill for building the network. It sells gateway equipment, software and related services to carriers, and it completes milestones on US government contracts.² Revenue was $70.9M in 2025 and is $88.6M in nine months of 2026, with a third quarter of $42.3M. The company says it is on target for $150M to $200M for the year, which means the fourth quarter has to bring in $61M to $111M C (calculated).²³

The plan for revenue is a share of what carriers charge. A carrier might offer a day pass, a monthly add-on, a plan for places with no towers, or include the coverage in a premium plan. AST says it generally seeks a revenue-sharing model. It does not say what share, and it does not say what carriers will charge.³ That one gap holds up the whole valuation, and I come back to it in Section VI.

The company reports $1.3B of remaining performance obligations, which is work contracted but not yet recognised as revenue. Only 10.1% of it, about $130M, is due in the next twelve months. The figure leaves out the carriers’ revenue share, because the amount is uncertain.⁴ The balance sheet shows $263M of contract liabilities, which is money partners have paid for service not yet delivered C (calculated).²

The partners

The company says it has partnerships with over 60 carriers that together serve over 3 billion subscribers. Those are the company’s words and the number counts everyone a partner serves, not the people who will pay. The 10-K is stricter about what is signed: definitive commercial agreements with AT&T, Verizon, stc and Vodafone, and about 50 preliminary agreements and understandings that have to be replaced by definitive ones before service can start in each country.²³

The Verizon payment shows how slowly the contracts move. Verizon agreed in 2024 to pay $45M ahead for service once AST had certain regulatory approvals and a definitive agreement. The agreement came in October 2025 and the approvals on 22 April 2026. The 10-Q of 9 November says the payment is due from Verizon and does not say that it has been received.⁴

The latest quarter

Revenue in the third quarter was $42.3M against $14.7M a year earlier, and almost matched the $46.3M of the first six months together C (calculated). The net loss before the share belonging to outside holders of the operating company was $171.8M, or $0.44 a Class A share.² Cash used in operations was about $169M and capital spending about $295M C (calculated).²¹² Cash and restricted cash ended at $3.23B, up from $2.72B at the end of June, because $1.15B of convertible notes were sold in July.²¹³

The headline of the release is that nationwide intermittent service will switch on before year-end in the United States, Canada, Japan, Saudi Arabia and the UK, with 12 satellites in orbit. The 10-Q has a stricter test. It says the company needs 25 satellites, five Block 1 and 20 Block 2, for what it calls noncontinuous service, 45 to 60 for continuous service in the key markets, and about 90 for all the markets it targets.²⁴ I can’t reconcile 12 with 25, and the company doesn’t explain it. I read “intermittent” as coverage only while a satellite is overhead, which at 12 satellites would be brief and spaced out. I will take the first date on which a customer pays for the service as the real start.

The stock closed at $36.89 on 9 November, 14% above the day before, and the results came after the close: the 8-K was accepted at 4:49 pm New York time.¹¹⁴ On Friday 6 November, after the close, the company had announced the Firefly rocket and Rakuten joint venture, so the close reflects those and not the quarter.¹⁵¹⁴

Where AST is from

Abel Avellan founded the company in 2017. He had earlier built Emerging Markets Communications, a satellite services business for ships and other moving customers, and sold it to Global Eagle Entertainment for $550M in 2016.⁹ The company went public in April 2021 by merging with a blank-check company.¹⁶ It is based in Midland, Texas, and had 1,126 employees at the end of 2025.³ Its first revenue came from a parts business, Nano, that it sold in September 2022, and revenue was zero in 2023.¹⁷¹⁸ Everything since has been spent on building the network.

IIThe moat

AST has four kinds of protection. Each is real. None is a wall.

Spectrum

Phones only hear certain radio frequencies, and the right to use them is scarce. AST’s plan uses three sources. The first is the carriers’ own low-band spectrum, at 700 and 800 MHz, which AST uses under agreements with Verizon, AT&T and FirstNet. The FCC order of 21 April 2026, DA 26-391, lets AST operate 248 satellites, 223 of them new, and serve phones on those carrier bands, 700 MHz and 800 MHz, in the contiguous United States and Hawaii. It comes with dates and a penalty. AST must launch half, 124 satellites, by 2 August 2030 and the rest by 2 August 2033. A missed date may cut the authorisation to the number of satellites then in use, and AST must report to the FCC by 17 August 2030 on the first date.⁶ The second is mid-band spectrum from Ligado, up to 45 MHz in the United States and Canada. That deal is not closed. It needs regulatory approval, and a fight with Inmarsat over regulatory support is still in the courts. AST has paid $520M so far, of which $100M sits in escrow by court order, and a backstop from Ligado’s sponsors is meant to refund it if approval fails.⁴ The third is priority rights to 60 MHz of S-band spectrum, got by buying a UK company in September 2025, for use in low orbit worldwide.³

The company aims for about 100 MHz in the United States and more than 60 MHz elsewhere. It reports a commercial authorisation in Brazil covering a combined 50 MHz of S-band and low-band spectrum.² The Motley Fool reported on 13 October that Deutsche Bank’s analyst Bryan Kraft says spectrum backs all but about $10 of the share price. That is secondhand, I can’t check it, and it was said when the stock was higher.¹⁹

What spectrum costs is easy to see next door. SpaceX, which has its own page on this site, is paying about $19.6B in shares and cash to EchoStar for 50 MHz of AWS-4 and H-block spectrum and up to 15 MHz more.⁸²⁰ That is the price of owned spectrum and it counts against AST in two ways. It makes AST’s rights look cheap. It also gives the main rival the same kind of asset.

Patents and know-how

AST reports about 3,500 patent and patent-pending claims, of which about 2,100 are granted or allowed, in 40 families. The first expire in 2039.⁴ The company designed the satellite, the phased array building block it calls a micron, and its own chip, and it controls the making of about 95% of the Block 2 subsystems.³ That is hard to copy. The test is whether a rival has to copy it at all. A different design that reaches a phone can use smaller arrays and more satellites. SpaceX has done that.

Carrier contracts

Carriers who sign with AST get a service they don’t have to build. The filings give no exclusivity terms that I could find. Two US carriers, AT&T and Verizon, use AST, and the FCC order names both along with FirstNet. SpaceX’s partner in the United States is T-Mobile.³⁷ If the contracts have no exclusivity, a carrier can use both networks. If they do, the protection is stronger. I can’t tell which. A newer fact cuts the other way. AT&T’s 10-Q says that on 14 September 2026 it agreed with Verizon and T-Mobile, SpaceX’s carrier, to form a joint venture that pools certain spectrum so satellite providers can reach more customers “through a unified platform”, and that AT&T will give access to spectrum to providers who contract with it.²¹ AST’s two US carriers and SpaceX’s now sit in one venture. I don’t know whether AST sells through it or beside it.

Scale in making satellites

AST has built the space for assembling and testing six satellites a month and says orders for materials cover the large majority of a 90-satellite constellation.⁴ Making dozens of the biggest commercial arrays ever flown is a skill that is not common. It is also a skill that has so far produced 8 launched Block 2 satellites since December 2025, 7 of them working.⁴

What weakens it

Three things. First, a rival is ahead in numbers: SpaceX has about 650 mobile satellites, about 7.4M monthly unique devices on about 30 carriers, and plans bigger satellites on Starship from 2027.⁷ Second, AST depends on others for launch. As the Motley Fool reported it, Piper Sandler’s Alexander Potter lowered his rating on 13 October because AST relies on SpaceX, its main competitor, to launch; three satellites wait at Cape Canaveral for a SpaceX date.¹⁹⁵ Third, a licence does not stop anyone using other spectrum.

My reading: AST has a head start in the capacity of a single satellite and a set of carrier and spectrum agreements that took years to sign. That makes a narrow protection that has not been tested. Rivals include Iridium, Globalstar, Skylo, Inmarsat and Thuraya, which the 10-K names, and above all SpaceX.³

IIIThe competitive advantage

The first section said where AST is protected. This one is about what it does better than the people it competes with, and where it does worse.

What AST does better

A bigger antenna on each satellite. The company says its Block 2 arrays are the largest ever flown in low orbit for commercial use. A bigger antenna makes narrower beams, which re-use the same radio frequencies more often and put more signal on the phone. In the 10-K the design target with the new chip is 120 Mbps peak on a beam with 40 MHz of spectrum and up to 10,000 MHz of processing bandwidth per satellite. The Block 1 satellites have shown a peak of 98.9 Mbps to an unmodified phone.⁴³ Those are company numbers from tests. I haven’t seen an independent measurement of what a phone gets at a busy hour, and that is the number that decides the revenue.

Broadband, not only texts. SpaceX’s service today is texts, light data and voice through apps such as WhatsApp. AST aims at voice, video and data on the carrier’s own network, with the satellite handing off to the tower and back. AST recently validated that hand-off with AT&T.⁷² If broadband to a phone matters to customers, this is the difference. If texts and an SOS button are enough for most people, then SpaceX’s simpler offer is enough.

More than one carrier per country. In the United States AST has AT&T and Verizon, the two largest, so each has a reason to fear the other will have the coverage first. The carrier venture described in Section II may change that.³

A second customer, the government. The satellites can serve defence work as well as phones. AST has 16 US government awards, a roughly $100M Space Force contract for sensing, a US priority rating, called DPAS, for its dual-use programme, and a spot in the Missile Defense Agency’s framework.²³ The same Space Force programme gave Rocket Lab $397M; see its page.²³ The government revenue is a smaller prize than the phone market, but it pays on milestones, and it has already paid.

Where it does worse

Launch. SpaceX launches its own satellites on its own rockets: its page here notes that 74% of its Falcon launches in 2026 carried its own satellites.²⁰ AST depends on SpaceX, Blue Origin and others. A New Glenn flight lost BlueBird 7. The company’s answer is Blacklift, a 20-ton reusable rocket to fly up to five BlueBirds at a time from late 2028, using engines licensed from Firefly. It costs $400M in cash and 10M to 14M shares to license the engines, then $250M for services and $250M for the first three sets of engines.²⁴ I can’t judge whether a rocket can be designed, built and flown in two years. The best comparison on this site is Neutron, the medium rocket of Rocket Lab, announced in 2021, then expected in 2024, then 2025, then 2026, and now 2027.²³

Cost per satellite. AST puts the average capital cost for a 90-satellite constellation at $21M to $23M a satellite, higher for the first ones.⁴ SpaceX does not publish a cost, but it makes the rocket, the satellite and the ground gear itself.²⁰ I don’t know whose cost for each unit of capacity is lower.

Time. SpaceX has T-Mobile as a customer and about 650 mobile satellites at 31 March, more than 50 times AST’s 12 today. A lead of a few years may matter more than any difference in the antenna.⁷ Deutsche Bank’s analyst argued in October, as TIKR reported it, that AST is not targeting consumers directly while SpaceX plans to use its new spectrum mainly for that, and that T-Mobile might move its business to AST; I treat that as an opinion and not a fact.²⁴

IVManagement

Who runs it

Abel Avellan founded the company and is its chairman and chief executive. Andrew Johnson, the chief financial and legal officer, joined in May 2024 from 3D Systems, where he had been chief legal officer. Dr Huiwen Yao, the chief technology officer, joined in 2018 from Northrop Grumman’s space business.⁹ The people building the satellites have been there a long time. The finance team is newer.

How they are paid, and what it pays them to do

Mr Avellan takes no salary and no bonus. His 2025 pay was stock awards of $14.2M. The other four executives have a $250,000 salary and a bonus, but nearly all of their pay is stock: $8.8M of $9.6M for three of them.⁹ So everyone is paid in shares of a company that has needed to issue them. That is not a fault. It is what a young company without cash profit does. It does mean the people running the company gain when the price is high and when the company raises money, and the pay plan counts both.

The 2025 performance share awards had eleven goals. Four of them are about money raised, held or funded: closing and financing the spectrum deals, keeping a level of liquid assets, the level of capital raised, and the number of satellites fully funded. A fifth is the average share price for the year. Others count satellites in orbit and built. The revenue goal was $75M. Revenue came in at $70.9M and the bonus paid 95% of target.⁹ The scoring of the awards was uneven: Mr Avellan earned 73.75%, Mr Wisniewski 98.75%, Mr Johnson 80% and Ms Gupta 43.75% of target.⁹

One thing needs saying. In May 2025 the compensation committee changed the goals of a 2024 award for three executives to match “shifting performance priorities”, with the new goals to be met by 15 August 2025. In August it found that 150% of target had been earned for Mr Wisniewski and Mr Johnson.⁹ On 25 September 2026 the committee adopted a policy that pays a lump sum on a firing after a takeover: two times salary plus target bonus for the chief executive and one and a half times for the others. It also turns unearned performance awards into time-based ones, unless the deal price is under the share price a goal needed.²⁵

How much of their own money is in

Mr Avellan owns 78,163,078 Class C shares, which carry 71.6% of the votes at the record date in April and are tied to the same number of units in the operating company. That is 20% of the economics, worth $2.9B at $36.89.⁹⁴ His shares are the ones that matter, and he made no open-market sale in 2026: his Forms 4 show awards and tax withholding only.¹⁰ Against that, 2,500,000 of his shares are pledged under a prepaid forward contract, a loan-like deal that he says did not involve a sale and does not change his votes or board rights.⁹ It is about 3% of his holding.

The other officers sold; the table below has each trade, and Mr Johnson has 49,985 shares pledged.¹⁰ Together the sales are $24.1M from four officers and one director, against $9.6M of pay for each of three of them last year C (calculated). The only open-market purchase in 2026 was by director Adriana Cisneros, 10,822 shares for $0.6M at $57 to $59 in late August.²⁶

Rakuten’s founder, Hiroshi Mikitani, reported selling 3,040,000 shares on 14 and 15 April at $86.22 to $91.42, about $271M, leaving 27,980,155 shares reported.²⁷ Rakuten is now a joint-venture partner with a 51% stake in RAST and was a holder of 21.0M shares, 7.0%, at the April record date.⁹⁴ I don’t know why he sold and I won’t guess.

Who else owns it, and who controls it

The control sits with Mr Avellan. His Class C shares carry up to ten votes each, set so that he and his permitted transferees control about 71.6% of the vote. The company calls itself a controlled company under Nasdaq’s rules. Under the stockholders’ agreement he can nominate seven of the 11 directors, and the number falls as his voting power falls.⁹ Other holders over 5% at the April record date were Vanguard with 7.2% of Class A (it later reported its holdings by division) and Rakuten Mobile with 7.0%. Vodafone Ventures has 1.8% of Class A and 80.6% of Class B.⁹ Alphabet Inc.’s Form 13F-HR for the quarter ended 30 September 2026, filed on 6 November, lists 8,943,486 AST SpaceMobile Class A shares held through Google LLC, worth $526.4M at that quarter-end. The same share count is in its March and June filings.²⁸

What they did with the money over a cycle

Since 2022 the company has raised about $2.3B from shares and warrants and about $5.0B from debt, almost all of it convertible notes C (calculated).^17291834 The share count rose from 181.5M at the end of 2021 to 390.0M at the end of September 2026, up 115%. It also paid a price to avoid cash costs. When it bought back $456.5M of its 4.25% notes and $250M of its 2.375% notes with new shares, it booked a non-cash cost of $99.7M in 2025 and $88.7M in nine months of 2026 for inducing holders to convert R (reported).³⁴ That is $188M for the right to pay in shares. The company has also bought capped calls, a hedge on its own convertible notes, at a cost of $98.6M in 2025 and $111.4M in July 2026, and sold the first set back in November 2025 for $74.5M.³⁴

The capital went into satellites, ground sites, spectrum and the Ligado deal. Capital spending was $1,065M in 2025 and $1,154M in nine months of 2026, and $520M went to Ligado.³² The test of any of this is the track record on promises, and it is the weakest part.

The company has done hard things. It flew five large satellites, then eight of a much larger kind. It got the FCC to license 248 satellites. It built the first factory for them and signed carriers and a government. It has not done what it said when it said it. A plan for 45 to 60 satellites by the end of 2026 has become, in November, a plan for about 45 by mid-2027, with 12 in orbit. I treat every date in the company’s releases as the earliest possible and my base case as a year later.

Lawsuits

A securities class action filed on 14 September 2026 in the Western District of Texas names the company and some executives. It covers buyers of the stock from 4 March 2025 to 16 July 2026 and alleges false or misleading statements. Two shareholder suits on the company’s behalf were filed on 21 October and 5 November 2026 against current and former executives and directors. The company says it will defend them. I can’t judge the claims, and they are early.⁴ The stock fell 17% on 16 July 2026, the day after the company announced its convertible notes offering, and that is the last day of the class period. I don’t know what the suit says about it.¹¹³

VFinancials

AST has reported six full years under its name, 2020 to 2025, and three quarters of 2026. There is no ten-year record to show. The record it has is one story: revenue small and unrelated to the plan, losses that keep growing, and from 2025 capital spending that is several times everything before it.

The loss is the smaller number. The net loss was $721M in nine months of 2026, and the cash an owner had to find was $1.78B, 2.5 times as much. The loss leaves out the satellites, because capital spending goes onto the balance sheet and is charged as depreciation over seven years. It also includes some costs that are not cash: $171M of stock pay and $89M for inducing holders of the 4.25% and 2.375% notes to take shares. I count stock pay as a cost, because shares are issued to pay it.²³

The running costs

The cost of running the company has doubled in a year. Adjusted operating expenses, which leave out stock pay, depreciation and the lost satellite, were $133.6M in Q3 2026 against $67.7M in Q3 2025. Without the cost of the gateways and government work, they were $109.4M, up from $95.9M in Q2.²⁵ That stock pay is the reason for much of the rise. It was $171.2M in nine months of 2026 against $32.3M a year earlier.² The fall from Q2 to Q3 in the loss comes from the lost-satellite charge in Q2, not from lower costs.

How the money was raised

The company paid for this with other people’s money. In 2025 it raised $1.30B from shares and $2.61B from debt. In nine months of 2026 it raised $0.09B from shares and $2.20B from debt.³² The debt is almost all convertible notes: holders lend at a low coupon, 1.625% to 2.25% on the new ones, and can swap the notes for shares at a price far above the stock’s today. That is cheap money, and its price is paid in the future share count.

The notes show how the market judged the credit at the end of September, when the stock was $58.86. The 2036 2.25% notes were worth 78 cents on the dollar and the 2036 2.00% notes 85 cents.⁴ A convertible note usually trades near face when the share price is close to its conversion price; these were far from it. The stock has fallen 37% since. At $36.89 none of the four large issues converts: the nearest conversion price, $72.07, is 95% above the price. If every holder did convert, 40.3M shares would be added, 10% more than there are today C (calculated).⁴ I treat the notes as debt, and I show in the valuation what happens to the per-share value if the stock rises past the conversion prices.

Since 2021 the share count has risen from 181.5M to 390.0M. Add 13.0M in awards that may vest and 12M for Firefly, the middle of the 10M to 14M promised, and the count I use in the valuation is 415.1M, before the 40.3M shares behind the convertible notes.⁴ A second class of holder sits in that count. Classes B and C are units in the operating company, AST LLC, that can be swapped for Class A shares. They are 22.9% of the company, 11.2M held by Vodafone and others and 78.2M held by Mr Avellan, and I add them to the count and do not treat them as a separate claim on value.⁴

Does the money last?

The 10-Q says cash on hand is enough for the next twelve months. It also says the company believes it is fully funded to make and launch about 90 satellites.⁴ The first claim I can check. The second is harder, because the 10-Q was signed three days after the Firefly and RAST agreements and does not say whether their cash is inside it. The table below puts the known calls on cash side by side.

The sources are cash of $2.8B and some hopes. The uses already agreed add up to a good deal more: the Firefly payments of about $0.9B in cash, the satellites, the spectrum payments and about $0.5B a year of running costs. The $1.4B for RAST needs a word. RAST uses its money to buy satellites, launch support and software from AST, so much of what AST puts in comes back as revenue, and AST may use the spare capacity.⁴ I can’t net that out, so my model leaves out both the $1.4B and the capacity. In my base case, before RAST, the cash left at the end of 2027 is about $0.8B and the company needs about $0.9B of new money by 2029 E (estimate). If the whole $1.4B were a plain cost, the need would be about $2.4B. The valuation shows this in a chart. The company has raised money every year, and the 10-Q makes building beyond the funded 90 satellites depend on raising more.⁴

The price is the market’s verdict on the raising. It reached $133.09 on 28 May 2026 and has closed as low as $30.20, 77% below the high. Over the same months the company sold $1.15B of 2034 notes. On 16 July, the day after the announcement, the stock fell 17%.¹¹³ The price on 9 November, $36.89, is 72% below the high. It was up 14% on the day, after Friday’s news of the rocket and the Rakuten joint venture, and before the quarter was released.¹¹⁵¹⁴

VIValuation

First, how this goes badly wrong

Before the numbers I asked what is most likely to hurt an owner here, and worked back from it. Three things stand out.

The service earns less than the network costs. On my cost figures the equity starts to be worth something only when the network takes in about $2.5B a year from the base satellites, which is 52 million subscribers at $4 a month. Below that, everything the service earns goes to debt, satellites that wear out in seven years and the cost of running the company. The base case has 70 million. If it is 20% lower, at 56 million, the base value falls from about $11 to about $3. If carriers pay $2 and not $4 a month, the value is nothing E (estimate).

The money runs out first. Cash is $2.8B, and I expect the fourth quarter alone to use about $0.95B of it E (estimate), including the $400M Firefly payment. In my base case the company has about $0.8B left at the end of 2027 and is short by about $0.9B at the low point in 2029 E (estimate). That leaves out the $1.4B for the Rakuten venture; counted as a plain cost, the shortfall would be about $2.4B. The risk is the price. Raising $0.9B at $20 a share costs 47 million new shares, 12% of today’s count. Raised at $36.89 it costs 25 million, 7%. If the market is closed, the company cuts spending and slows the build, which is the bear case.

The build takes another year, then another. The company has moved its dates twice in two quarters. Pushing the whole base case back one more year, with the same cash spent in the same years, takes the base from about $11 to about $7, and the same delay takes the bull from about $39 to about $33 E (estimate). A rival that takes the best carriers does the same damage without a delay.

None of these needs anything exotic. They set the bear case, and they are the reason I start from cash that reaches an owner and not from earnings, which are negative.

What an owner could really take out

I value cash after stock pay and after the capital needed just to stand still. For AST that second part is large. A satellite lasts about seven years, which is the life the company uses for depreciation, so a network of 125 needs about 18 new satellites a year just to keep going, at $18M each, plus ground equipment.⁴ The table starts from what a year of nine months looks like now, then gives my base for two later years.

In nine months of 2026 the owner cash was −$1.78B against a net loss of $0.72B C (calculated). In my base for 2035 it is $1,150M a year against a net income of about $1,303M E (estimate), because by then replacing satellites costs more each year than the depreciation on them.

What the price implies

Put the other way, what does $36.89 need? At a 12% discount rate the price needs about 97 million paying subscribers at $5 a month, once the full 248 satellites are up. That is $5.8B of service revenue a year, or 6.7 times the whole of Iridium’s revenue, from 3.2% of the 3 billion subscribers its carriers serve. On the base satellites the price needs 117 million at $4. Starlink Mobile, the closest thing now running, reached about 7.4 million monthly devices by 31 March 2026, with about 650 satellites and more than two years of service.⁷²²

What I can’t judge

• What a carrier will pay per subscriber. The filings say revenue will be shared and give no price. SpaceX charges its carriers a fixed fee or a fee per user and does not disclose either.⁷ My $3 to $5 a month is a range and not a finding. The valuation is more sensitive to it than to any other input.
• How many users actually pay. Starlink Mobile counts monthly devices, and many of those get the service as part of a plan. A paying subscriber is not a device. I use paying subscribers and I don’t know the conversion.
• What one satellite can carry at a busy hour. The design numbers are the company’s, and a satellite that serves a whole country at 6 pm is a different thing from one that makes a good demonstration.
• Launches and the cost of reaching orbit. Ten launches are contracted. Blacklift is a promise for late 2028. I use the Neutron history as a guide.²³
• What SpaceX does with its spectrum. The EchoStar deal, up to 65 MHz, is for direct service. It might take the carriers that matter most, or leave room for AST.⁸

Because I can’t know these, I show a range and where the price sits in it, and I don’t treat $11 as exact. A fair reading of my work is that the base is somewhere from nothing to about $25, depending mainly on subscribers and price per subscriber.

The discount rate

I ask 12.0%. The 10-year Treasury yielded 4.88% on 9 November and the implied equity risk premium on 1 November was 4.41%.³²³³ A beta of 1.6 gives 12%. The stock’s own beta is 2.9 on the quote I read, which gives 17.7%.³⁴ I don’t believe that figure measures the risk of the business, since it comes from a stock that has ranged between $27 and $133 in a year. I do believe the business is riskier than the 10% I use for steadier companies on this site. SpaceX is at 10.5% and Rocket Lab at 13%.²⁰²³ At 9% my base is $23 and at 14% it is $6. The bull is $68 at 9%.

Three cases

Bear

Launches keep slipping and the money gets dearer. The company reaches about 55 satellites by 2031 and stops. Carriers like the product and few customers pay: 12 million subscribers at $3.50 a month, or $0.5B a year. Running costs are cut after 2028 and it still does not cover $4.2B of debt and $0.9B of rocket payments. The shares are worth nothing, or an option on a recovery. I show $0 and say why below.

Base

The company launches about a year later than it says: 35 Block 2 satellites by the end of 2027, 100 by 2030 and 125 by 2033, with one in 17 lost. That is about 100 satellites in orbit in August 2030 with Block 1, against the 124 the FCC order asks for, so the base counts on the FCC moving that date or not enforcing it. Continuous service in the key markets starts in 2029. Carriers pay AST $4 a month for 70 million subscribers by 2033, 2.3% of the 3 billion their carriers serve, or $3.4B a year. It needs about $0.9B of new money at the low point in 2029. Value $11.

Bull

The company launches close to its plan: 45 Block 2 satellites in 2027, 90 in 2028 and the full 248 the FCC licensed by 2033 (the five Block 1 satellites are inside that 248, so I assume they are retired by then), helped by Blacklift after 2029. A hundred million paying subscribers at $5 a month give $6.0B a year. SpaceX wins its own market and AST wins the carriers that want broadband. Value $39, a little above the price.

The model is one page of arithmetic and I kept it in my working folder; the method block at the end gives its steps. I assume half the cost of the next 36 satellites is already paid, because the 10-Q says supplier orders are in place for the large majority of a 90-satellite constellation and microns are built for satellites up to BlueBird 51.⁴ I leave out the $1.4B subscription to RAST and also the mid-band capacity it buys, because RAST pays AST for most of what AST puts in and I can’t net the two. If I count the $1.4B as a plain cost with nothing in return, the base falls from $11 to $8 and the bull from $39 to $36. Stock pay is $180M in 2027, under the third-quarter pace of about $210M a year; at $210M the base is $10.

Other uses of the same dollar

The 10-year Treasury pays 4.88% with no launch.³² My base cash flows return 7.3% a year at $36.89, which is 2.4 points above the Treasury and 4.7 points short of the 12.0% I ask. My bull returns 12.3%, just above what I ask. The best comparisons on this site are the two other space companies. The price of SpaceX is 182% above my base and that of Rocket Lab is 465% above mine, so AST’s 235% sits between them.²⁰²³ SpaceX has $32.0B of revenue and the strongest position in launch. Rocket Lab has $769M of revenue and an order book. AST has $143M and a plan. By my numbers none of the three clears its bar at its price, and AST’s gap is the one that depends most on a single input.

Why the bear case is $0

The bear is what my numbers give when the service earns little. In that case the model says −$18 a share, and shares cannot be worth less than nothing, so I show $0. Three things make it so. The network never pays for itself: 12 million subscribers bring in $0.5B a year, and the company spends more than it earns in every year to 2040 E (estimate). The money has to come from somewhere: cash is gone in 2028 and the hole is $1.6B by 2030. And the notes do not help shareholders. The large issues convert at $72 to $116, so at a low price they stay as debt, $3.7B of it due from 2032 to 2036. To fill the 2030 hole at an illustrative $10 a share would take 163 million new shares, 42% more than today. The other way out is to restructure the debt, and then the shares are paid only after the creditors, with perhaps nothing left. I give this case 45%, and that is a guess. The $0 is a floor and not a forecast for each share: a little value could survive as an option on a recovery, and I don’t count it.

The margin of safety

I set it from the bear case: a price at which my bear value is at least three quarters of what I pay. My bear value is $0, so that price is $0. No price above zero has a margin of safety here, which is what a business with $4.2B of debt and no revenue from its main product looks like in a bad year. I am not saying the shares will go to nothing. I am saying a buyer pays for the possibility that they could, and gets no cushion for it.

Two different questions

The first is how good the business is. If it works, it is a very good one: a wholesale network with high margins, carriers as customers and a growth market. The numbers I use for 2035 are a business with operating margins of about 60% before stock pay. The second question is how good the price is. At $36.89 the price is 235% above my base value and 5% below my bull, which assumes the plan is met and carriers pay for 100 million subscribers. A good business at a price that already assumes it works leaves a buyer exposed to every delay, and AST has delayed more than once. I give the range and the price, and the reader decides.

VIIThe case for the price

What most people miss

Three facts cut in AST’s favour. Partners have already put in about $218M: the balance sheet carries $263M of contract liabilities for service not yet delivered, and $45.5M of that is a payment still due, most likely Verizon’s C (calculated).⁴ AST did not have to buy its main US spectrum. It uses Verizon’s, AT&T’s and FirstNet’s under lease, and the FCC has licensed 248 satellites on it, while SpaceX is paying about $19.6B for EchoStar’s.³⁶⁸ And Japan will pay for part of the build: its grant to the Rakuten and AST venture is up to ¥148B, about $938M, roughly 40% of the cost of the venture’s satellites C (calculated).⁴

The bull case

Mr Avellan’s case, in the release of 9 November, is that AST becomes the direct-to-phone partner of the carriers and the US government. He expects to work with all the major carriers in the United States and Europe, production to reach six satellites a month in December, Blacklift to give control of launches, and 2026 revenue to stay on target for $150M to $200M.² As TIKR reported it, Deutsche Bank’s Bryan Kraft raised his rating on 12 October and kept a $93 price target. He called the fall after SpaceX’s spectrum purchase misguided, said AST does not target consumers while SpaceX does, and said T-Mobile might move its satellite business to AST.²⁴ Nasdaq’s one-year target field, a data-feed figure, read $78.00 on 9 November.³⁵

The next day, as the Motley Fool reported it, Piper Sandler’s Alexander Potter lowered his rating, saying SpaceX is now AST’s main competitor and AST depends on SpaceX rockets to launch.¹⁹ Two firms read the same company and disagreed within a day. I have not opened either note. The bull case, in plain words: carriers want one broadband network they can plug into their own systems, and a few billion dollars a year of wholesale revenue is the prize.

Against reality

If AST launches about as fast as it says, carriers pay for 100 million subscribers at about $5 a month, and the money keeps coming, the price is possible. That is my bull, $39, and the price is 5% below it. That is 7 Block 2 satellites to 248 in seven years, no paying subscribers to 97 million, and $143M of revenue to about $6B. The market may be right: satellite coverage could become a normal part of a mobile plan, as roaming did, and AST has the licence and the carriers.

Multiplying the price out, the bull needs five things: money on fair terms, about 80%; about 120 satellites by 2029 and 248 by 2033, about 30%; broadband at a busy hour as designed, 65%; carriers paying for 100 million subscribers, 30%; and SpaceX, with 65 MHz of its own spectrum, leaving the best carriers alone, 55%. Multiplied, that is about 3%, or one chance in 39. They are guesses and linked, because if the satellites work, carriers will sign. I give the bull 15% in my weights, the softest number on this page E (estimate).

What a 10x would take. Ten times $36.89 is $369, or $144B of value on today’s 390.1M shares. At 20 times net income that needs $7.2B of profit a year. At a 50% net margin that is $14.4B of revenue, 101 times the last twelve months and about as much as Starlink’s whole Connectivity business earns now, $16.0B. Done in seven years, it is 93% growth a year, every year, with $4.2B of debt to carry E (estimate).²⁰² My bull is $39, so $369 is 9.5 times my bull. For most large companies a tenfold move is close to impossible. For one with this much still to build it is not impossible, but I can’t name a company that has grown revenue a hundredfold in seven years at this scale, and the price already holds most of the bull case.

Where I come down

Three things moved me: the Japanese grant, the 248-satellite licence and the pace of the factory. Two did not: a record of late dates and the missing price per subscriber. The third-quarter revenue is gateway and government work, so it moved my 2026 estimate and not my base. What would raise my values is evidence on the one number I can’t judge: a paying count above 8 million by the end of 2028 at $4 a month or more to AST, with 65 satellites up and the next $0.9B raised at $37 or better. What would lower them is a fourth date moved, a failed launch, the Ligado deal failing, or a carrier venture that leaves AST out.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Launch. A rocket put BlueBird 7 in too low an orbit. Ten launches are contracted and at least five more are agreed but not binding. The two rockets AST relies on most are SpaceX’s and Blue Origin’s.¹¹⁴¹⁹
• The chip. Production moves from programmable chips to AST’s own, the AST5000, from BlueBird 34. A change like this late in a build can stop a line.⁵⁴
• Regulators, country by country. The FCC licence is in hand, with a dated condition: half the satellites by 2 August 2030, or the licence may shrink to the number in use.⁶ Each other country needs its own approval, and the company says it needs definitive agreements with the carriers that replace the preliminary ones before it can start service there.³ Brazil has authorised commercial service with a combined 50 MHz.²
• Spectrum. The Ligado deal needs regulatory approval and survived one court fight with Inmarsat, which is appealing. If it fails the $520M is meant to come back through the backstop, and the plan loses 45 MHz in the United States and Canada.⁴
• Currency. The Japanese grant is in yen. The company used 157.75 yen to the dollar for its $938M. A stronger dollar makes the grant worth less.⁴
• Carriers. AT&T and Verizon, AST’s two US customers, have agreed with T-Mobile, SpaceX’s customer, to pool spectrum in a joint venture that satellite providers can contract with. If the venture favours SpaceX, AST’s US position is weaker than its two agreements suggest.²¹
• One person. Mr Avellan controls 71.6% of the votes, nominates seven of 11 directors, and the company depends on one founder. A controlled company is exempt from some Nasdaq governance rules.⁹
• Markets. The notes fall due from 2032 to 2036, and it paid $41M of cash interest in nine months of 2026.⁴ If the stock is low when they come due, repayment in cash or in shares both hurt.