Open in interactive viewer → charts, metric popovers & call review

Firefly Just Discovered It's an Engine Company, Not a Rocket Company

A $1.3 billion AST SpaceMobile IP license rescues the narrative while Alpha Flight 8 slips a third time — and nobody in the tape seems to mind
FLY · Earnings Call · 2026-11-10

The product was never the rocket

Firefly spent nine years being valued as a launch company. This quarter it quietly re-rated as an engine-licensing house. The AST agreement — more than $1.3 billion, explicitly non-dilutive — hands AST SpaceMobile the intellectual property behind the Miranda and Vira engines: roughly $800 million in cash and AST Class A stock ($400 million cash plus 10-14 million shares struck off a forward VWAP), $250 million of support services, and $250 million for engines to support AST's first three flights. The hardware bill of materials is 27 Miranda first-stage engines and 9 Vira upper stages — three complete flight sets for AST's own three-stage, 20-metric-ton-to-LEO vehicle.

When a company sets out to build its own rocket, the engine is the hardest, highest stakes decision they make. AST chose Firefly, and that tells you everything you need to know about where our technology stands.

Jason Kim, Chief Executive Officer · 2026-11-10
The structure matters more than the headline. Firefly keeps ownership of the IP; AST independently configures the engines for its own vehicle. And the counterparty risk is, on the CFO's telling, not Firefly's problem — a striking claim given that rocket programs historically run years late. “if there is any delay on their side, I don't believe it has any impact on us in terms of revenue recognition.” — Darren Ma, Chief Financial Officer · 2026-11-10 What the deal does not do is arrive quickly. The IP and services revenue is percentage-of-completion over three years; engine deliveries begin ramping in year two. Firefly booked $4.7 million of launch revenue in Q3 against $106 million of spacecraft revenue. Next year, that line item starts becoming the interesting one.

The slip the market chose to forgive

Alpha Flight 8 — the debut of Block II — moved again. It was “late summer” on the Q1 call, and by the August call the analyst was already tracking the drift: “You noted Alpha 8 into Q4 from late summer and 10 to '27.” — Sheila Kahyaoglu, Analyst · 2026-08-11 Jason Kim said then that “We are going to have several Alphas in this year.” — Jason Kim, Chief Executive Officer · 2026-08-11 Now the answer is zero Alphas in calendar 2026. Management blames a national security customer's manifest readjustment rather than hardware: “We did not make this decision lightly as our preference was to launch before the end of the year.” — Jason Kim, Chief Executive Officer · 2026-11-10 The factory evidence is genuinely thick — structures for Flights 9 through 12 in various states, 24 Reaver chambers in flow, four separate hot fires in a single three-hour window — so this reads as customer-and-range scheduling, not a repeat of the 2025 test-stand failure that a Morgan Stanley analyst pushed on last November. Still, it is the third consecutive quarter in which the launch-cadence promise has moved right, and the stock has already paid: down more than half from its August 2025 peak of $60.35. Which is exactly why the guidance raise was so modest — the low end went from $420 million to $440 million, midpoint to $445 million. Launch cadence barely touches this P&L anymore.

Margins improving, cash still leaving

Q3 GAAP gross margin was 22.9% versus 20.3% in Q2, on revenue of $110.7 million (+259% year over year) — a second straight quarter above $100 million, with year-to-date revenue of $309.2 million, up 203%. Backlog hit a record $1.7 billion; more than $300 million of new contracts more than replenished what the quarter converted. The reported fundamentals through June tell the same shape: total revenue of $118 million in the June quarter, up 657% year over year, against gross profit of $24 million. The offset is the spend side. R&D of $72 million against $118 million of revenue — roughly 61 cents of research per revenue dollar. That is the honest read on the AST windfall: it is not free money, it is pre-funding for a capex and engineering ramp that will show up first on the outflow line. Free cash flow was an outflow of $123 million in the June quarter, and the CFO was unusually direct about the direction of travel: “we could expect cash burn to tick up through next year as part of this contract.” — Darren Ma, Chief Financial Officer · 2026-11-10 Q3 free cash flow improved to a $85.3 million outflow against liquidity of $839.7 million — a $534.7 million cash pile plus an undrawn $305 million revolver. Enough runway. Not infinite.

Leaning into narratives the tape is de-rating

The moon franchise got structurally better this quarter: Firefly onboarded to the $10 billion CLPS 2 vehicle, which opens larger cargo landers and light terrain vehicles, and it is running five lunar missions in parallel out of a clean room that quadrupled in square footage. That is real optionality on a program whose scope is bigger than the $30 billion NASA has discussed for Moon Base. But two things deserve a raised eyebrow. First, the company's own data center narrative slipped down its keyword set even as Jason Kim talked up Starcloud flying an AI compute payload on Elytra to “validate the core capabilities of the future data center in the moon's orbit.” That is not company-specific caution — the entire high-bandwidth-memory and co-packaged-optics cluster is down double digits on the 90-day tape. Firefly is telling a space-compute story into a market that just stopped paying for compute stories. Second, space domain awareness — the phrase Firefly uses to straddle lunar and defense ambitions — appears in the global 90-day decliners alongside Lunar Outpost-adjacent names and Firefly itself. The company is expanding into a theme the tape is marking down. That is either a divergence worth owning or a warning; the Miranda engine qualification campaign (175-plus hot fires, 4,230 cumulative seconds) is the tangible asset anchoring the bullish case either way. The corroboration worth noting is cross-sectional: AST SpaceMobile reported the same week and its own keyword set included the Miranda engine and a launch agreement. When the counterparty independently flags the same technology on its own call, the licensing thesis is no longer just Firefly's slideware. The verdict: this was a genuine strategic pivot disguised as a contract win. Firefly stopped selling launches and started selling the thing launches are made of — and got paid in cash and a customer's equity for the privilege. What it did not do was launch a rocket this year.