Phase Space AI

Financial Model Notes

AST SpaceMobile [ASTS]

AST SpaceMobile, Inc. [ASTS] — Financial Model Notes

Every figure used in this memo, with its filing provenance and derivation. Nothing here is estimated unless the row says so.

Sources: XBRL companyfacts for CIK 0001780312 (fetched 2026-07-29); the primary HTML of the FY2025 10-K (2026-03-02), the Q1-2026 10-Q (2026-05-11), and the 8-Ks of 2026-07-15 and 2026-07-20; Alpaca SIP daily bars; Alpaca options snapshots.

---

1. Revenue — quarterly build and the TTM window

us-gaap:RevenueFromContractWithCustomerIncludingAssessedTax, USD.

TTM to 2026-03-31 = 1,156 + 14,739 + 54,305 + 14,735 = $84,935,000. Matches the screen exactly. Q4-2025 derived as FY minus 9M, which is correct and reconciles.

Concentration in time: Q4-2025 is 63.9% of the TTM, and Q1-2026 fell 72.9% sequentially from it.

A tagging note. The screen's cagr_base_revenue of $13,825,000 is RevenueFromContractWithCustomerExcludingAssessedTax for FY2024, whereas the FY2024 income statement reports $4,418,000 (...IncludingAssessedTax) and Revenues reports $4,400,000. Three different FY2024 revenue figures exist in the same XBRL set, spanning 3.1x. The income-statement figure of $4,418k is the one used here. Recorded because a CAGR base picked by tag preference rather than by statement reconciliation can be wrong by multiples.

1.1 Composition — the central finding

"To date, the Company has not recognized any revenues from its SpaceMobile Service." — FY2025 10-K

Product gross margin, Q1-2026: (13,400 − 11,100) / 13,400 = **17.2%**. Service gross margin: (1,300 − 600) / 1,300 = 53.8%. Blended: (14,735 − 11,700) / 14,735 = 20.6% — against the screen's 51.4%, which is the FY2025 blend struck on a very different mix.

Of FY2025 gateway revenue, $2.1m was to SatCo, a variable-interest entity accounted for by the equity method — a related party.

1.2 Remaining performance obligations

Excludes MNO revenue-share variable consideration, which is constrained under ASC 606 — i.e. excludes the investment thesis.

---

2. Income statement

ASTS presents no OperatingIncomeLoss line. All operating figures below are derived.

Operating loss = revenue − OperatingExpenses. OperatingExpenses as tagged includes cost of revenues; verified against the MD&A's component discussion (engineering services $84.1m + G&A $43.7m + R&D $7.1m + D&A $17.6m + cost of revenues $11.7m = $164.2m ≈ the tagged $164,147k).

Loss-allocation cross-check, which independently corroborates the share-count correction: 58,600 / (191,012 + 58,600) = 23.5% of the consolidated loss is attributed to the AST LLC minority. Class B plus Class C are (11,215,111 + 78,163,078) / 388,124,572 = **23.0%** of total shares. The two agree to 0.5pp, confirming that Class B and C carry economics and that the screen's Class A-only count of 290.7m is wrong.

2.1 EPS control — not runnable, and why that matters

ASTS tags no EarningsPerShareBasic or EarningsPerShareDiluted in the aggregated XBRL. The income statement presents "Net loss per share attributable to holders of Class A Common Stock", which is dimensionally tagged by class and does not survive SEC aggregation.

Recorded for the calibration file: on multi-class Up-C registrants the instructed net income ÷ shares ≈ filed EPS control has no data to work with and returns nothing — it does not fail loudly, it fails silently. The loss-allocation cross-check above is the substitute that worked, and it should be added to the control set: for any registrant with non-controlling interests, NCI share of loss ÷ total loss must reconcile to non-Class-A shares ÷ total shares. On ASTS that test catches the error the EPS test could not even attempt.

---

3. Balance sheet, 2026-03-31 (10-Q, $000)

THERE IS NO SHORT-TERM INVESTMENTS LINE. Total current assets of $3,152,599 = 3,029,591 + 873 + 27,453 + 16,756 + 10,673 + 67,253 exactly. The screen's $2,800,000k ShortTermInvestments cannot be accommodated anywhere in this balance sheet.

3.1 The $2.8bn phantom — root cause, verbatim

"As of March 31, 2026 and December 31, 2025, the Company had approximately $3.5 billion and $2.8 billion of cash and cash equivalents and restricted cash, respectively, of which approximately $2.8 billion and $2.0 billion, respectively, is classified as cash equivalents, which consisted primarily of short-term money market funds with original maturities of 90 days or less." — Q1-2026 10-Q, fair-value note

That sentence is the sole source of the $2,800,000,000 tag. It describes the composition of the cash already on the balance sheet. Note that the same sentence produced $2,000,000,000 at 2025-12-31, which the screen would also have added, so the error is not a one-off in this filing — it recurs each period.

Three independent tie-outs refute it, and all three were available: 1. Total current assets foot exactly without it (above). 2. The MD&A states liquidity as $3,458.9m including restricted cash — matching 3,029,591 + 873 + 428,400 = 3,458,864. 3. The MD&A's own words: "Our current sources of liquidity are cash and cash equivalents on hand" — no investment portfolio is claimed.

3.2 Net cash / net debt

Debt verification, as instructed. Checked LongTermDebt, LongTermDebtCurrent, LongTermDebtNoncurrent, ConvertibleDebtNoncurrent, SecuredLongTermDebt, DebtInstrumentFaceAmount, LongTermDebtFairValue, and the maturity schedule.

3,024,121 face − 52,589 unamortised issuance costs = 2,971,532 carrying ✓ the face, carrying and issuance-cost figures reconcile exactly.

Net cash at 2026-03-31, including restricted cash:
   3,029,591  cash and cash equivalents
         873  restricted cash, current
     428,400  restricted cash, non-current  [PLEDGED as 102% UBS collateral]
  (2,971,532) total debt, carrying
  -----------
     487,332  net cash  ->  $1.26 per share on 388.1m shares
  excluding the pledged restricted cash:  58,932  ->  $0.15 per share

Net debt now:
   2,723,000  cash + restricted cash at 2026-06-30 (PRELIMINARY, 8-K 2026-07-15)
  (3,024,121) debt face at 2026-03-31
  -----------
    (301,121) net debt at 2026-06-30
    +983,600  net proceeds, $1.0bn 2034 converts (8-K 2026-07-20)
     (96,900) capped call cost
  (1,000,000) new debt face
  -----------
    (414,421) NET DEBT used throughout this memo  ->  ($1.07) per share

Screen: +$2,866,295k. Verified: +$487,332k at the balance-sheet date, −$414,421k now. Error $2.38bn–$3.28bn.

---

4. The debt stack, instrument by instrument

Undrawn: Sound Point Credit Facility $550.0m, non-recourse to ASTS and AST LLC, SpectrumCo borrower, SOFR + 8.00% or ABR + 9.00%, available to 2026-10-05 (+180 days for a 1% fee), 2% commitment fee paid, 0.15% monthly ticking fee, 3% upfront fee on draw, 1%–5% termination fee, maturity 48–60 months from funding. Conditional on FCC and other approvals for the Spectrum Usage Rights Transaction.

No maturity before 2032. Repayment schedule at 2025-12-31: next 12 months $11,946k; year 2 $433,159k; year 3 $18,044k; year 4 $12,586k; year 5 $12,077k; after year 5 $1,776,623k.

4.1 Converts repurchased with equity — $918.6m of value transferred

$706.5m of face retired for ~$1,625.1m — a 2.3x premium, $918.6m above face — funded by issuing 23.6m shares to the note holders. Economically these were conversions executed at market on the deep-in-the-money $26.99 series. Q1-2026 carries $89.8m of induced-conversion expense for the February tranche.

---

5. Shares

Class B is non-economic at the registrant level but corresponds one-for-one to AST LLC common units redeemable into Class A (or cash at the Company's option); a corresponding Class B share is cancelled on redemption. Class C is held by the chairman and chief executive. Both are economically dilutive and both must be counted. No redemptions of AST LLC common units occurred in Q1-2026.

388,124,572 is used throughout this memo. The screen's 290,689,457 is the Q1-2026 Class A weighted average — a weighted average mistaken for an outstanding count, and two of three classes omitted. Undercount: 25.1%.

Equity issuance programmes. October 2025 ATM: $800.0m authorised across ten agents; 874,045 shares sold in Q1-2026 for $80.3m net ($0.4m of commission); "having utilized virtually the entire capacity", terminated 2026-03-17. Ligado penny warrants: 4,714,226 issued 2025-03-22 at $0.01, fully exercised in Q1-2026. Private placement warrants: expired 2026-04-06, final 122,000 exercised for 109,499 shares.

---

6. Capital expenditure and the funding arithmetic

Property and equipment, net: $1,398,761k (2025-12-31) → $1,638,262k (2026-03-31).

Company-stated programme cost (Q1-2026 10-Q):

Explicit exclusions and conditions in the company's own words: the per-satellite estimate "excludes cost of certain initial satellites used to validate satellite performance and operations" and is conditional on "securing future launch contracts with more favourable terms, diversifying our supply chain… economies of scale, continuous process improvements" — and "if we are unable to achieve" those, "the average capital cost… will be higher and such variations could be material."

Obligations the "fully funded" claim does not cover:

Conclusion recorded in ASTS_Research.md §3.1: the satellite manufacture-and-launch programme is funded on the company's own estimate; the company as a whole is not, and the company does not claim it is — two paragraphs later it writes "We plan to raise additional capital" and "Until such time, if ever, as we can generate substantial revenues to support our cost structure…". It then raised $1.0bn, 27 days after filing.

---

7. Reverse-DCF mechanics

Implementation identical to assets/reverse_dcf.py:

EV_target   = spot x shares - net_cash          (net_cash negative here, so EV = mkt cap + net debt)
EV_model(g) = revenue0 x (1+g)^5 x terminal_margin x exit_multiple / (1 + wacc)^5
solve g by bisection on [-0.50, +2.00];  multiple bounds [0.1, 200.0]
return None if the interval does not bracket the target  ->  "NO SOLUTION"

EV_target = 53.03 x 388.124572 + 414.4 = 20,584.3 + 414.4 = $20,998.7m

Central solvable cell (TM 15%, 18x, WACC 13%, 5y, revenue0 = 84.935):

  (1+g)^5 = 20,998.7 x 1.8424 / (84.935 x 2.70) = 38,687.9 / 229.32 = 168.71
  1+g = 168.71^0.2 = 2.7889   ->   g = 178.9%
  year-5 revenue = 84.935 x 168.71 = $14,329m ; EBIT at 15% = $2,149m ;
  exit EV at 18x = $38,688m ; discounted = $20,999m  OK
  implied year-5 EV/sales = 38,688 / 14,329 = 2.70x  vs 247.2x today = 91.6x compression

The no-solution boundary, verified:

TM 5%, 200x multiple, g at the +200% ceiling:
  EV_model = 84.935 x 3^5 x 0.05 x 200 / 1.8424 = 84.935 x 243 x 10 / 1.8424 = $112,051m

That is well above the $20,999m target, so at first sight a solution should exist — but the solver bisects on the CAGR at a FIXED multiple, and at the fixed multiples tested (12x–30x) the maximum achievable model EV at a 5% terminal margin is:

TM 5%, 30x multiple, g = +200%:  84.935 x 243 x 0.05 x 30 / 1.8424 = $16,808m  <  $20,999m

$16,808m < $20,999m at the CAGR ceiling, so the interval does not bracket the target and the solver correctly returns no solution. The same test at 12x gives $6,723m. This is a genuine no-solution, not a numerical artefact, and it is a no-solution in the CAGR at multiples of 12x–30x. Solving instead for the multiple at a 5% terminal margin returns no solution in 0.1x–200x at CAGRs of both 72.5% and 100% (§3.2 of ASTS_Valuation.md), which is the same finding approached from the other axis.

All grids in ASTS_Valuation.md §3 were produced by this routine.

---

8. Market data

Bars from 2021-01-04 (1,398 SIP sessions), split-adjusted. Option-implied volatility 99%–108% versus 108.9% realised — see ASTS_Trade_Construction.md §1.3.

---

9. Own-multiple history — construction and its limits

The first four rows use Class A weighted averages rather than quarter-end outstanding counts, because only the last two quarters' outstanding figures were extracted from the filed balance sheets. Those four rows are approximations and are marked as such. Net cash for the intermediate quarters is built from tagged cash less tagged debt and carries similar imprecision.

This does not matter, because the series is declared UNIDENTIFIED anyway (ASTS_Valuation.md §6): four of the six observations have TTM revenue between $4.4m and $18.7m, so the multiple is an artefact of a near-zero denominator. The table is shown for transparency, not for use.

---

10. Known gaps, stated rather than filled

1. Q2-2026 results are not available. Only preliminary cash of ~$2,723m at 2026-06-30 (8-K 2026-07-15), which the company explicitly labels as unaudited and incomplete. The ~$736m of quarterly consumption inferred from it is therefore preliminary, and its split between capex and operating expenses is unknown.
2. The 2036 2.00% notes' conversion price is not disclosed in the sections reviewed and is not asserted.
3. The Block 2 BB7 write-off is unquantified.
4. No Block 2 launch date, commercial-service date or Ligado closing date is disclosed, so none appears in the catalyst calendar.
5. Three different FY2024 revenue figures exist in the XBRL ($4,418k, $4,400k, $13,825k). The income-statement figure is used; the discrepancy is flagged, not resolved.
6. No consensus estimates — Alpha Vantage quota assumed exhausted. Consensus Criteria INDETERMINATE, blocks nothing.
7. No earnings-call transcripts — mention frequency computed on filing text, substitution declared in ASTS_Research.md §5.
8. No growth-matched peer comparator was constructed, and none can be for a company whose product revenue is zero. The exit multiple is UNIDENTIFIED by construction; the screen's exit_multiple: 28.5 from 63 peers is not relied upon.
9. WACC of 13.0% is an assumption, not a beta regression.
10. The own-multiple history's first four rows are approximations (§9).
11. No ASTS_Model.xlsx. Every derivation a workbook would contain is written out above and is reproducible by hand from the filed figures.