Relativity Space

Phase A — Understand the business

Company Overview

Relativity Space is a Long Beach, CA launch company founded in 2015 by Tim Ellis and Jordan Noone, originally built on a single radical thesis: 3D-print an entire orbital rocket. That thesis has been substantially walked back (see Lens 3), and the company today is best understood in two layers:

1. What it sells: launch services — payload-to-orbit contracts priced by mass and vehicle, plus a small line of additive-manufacturing parts work for customers like the US Air Force. Revenue is contract-driven and lumpy; the company books multi-launch Launch Services Agreements (LSAs) that convert to a backlog before any flight occurs.
2. What it is becoming: under Eric Schmidt (CEO since March 2025), the framing has shifted toward owning launch capacity for a future of orbital AI data centers — Schmidt confirmed publicly with a one-word "Yes" when asked whether he bought Relativity to put data centers in orbit.

The product. The flagship and now sole vehicle is Terran R — a two-stage, partially reusable medium-to-heavy lift rocket: ~270 ft tall, 18 ft diameter, 33,500 kg to LEO expendable / 23,500 kg reusable. First stage uses 13 Aeon R LOX/sub-cooled-methane gas-generator engines (~269,000 lbf each), all designed and built in-house. The earlier Terran 1 — the 85%-by-mass 3D-printed rocket — flew once in March 2023, survived Max-Q, but a second-stage LOX turbopump anomaly stopped it short of orbit. It was retired immediately after that single flight.

Customers (the backlog). Over $2.9B in Terran R launch contracts as of March 2025, up from ~$1.8B in late 2023. Named customers: OneWeb (>$1.2B / 20+ launches, signed June 2022), Intelsat (multi-launch, Oct 2023), SES (multi-launch, expanded Nov 2025 — the most recent vote of confidence), plus NASA (VADR), Iridium (largely lost to SpaceX), Impulse Space, and US government.

Contract structure & key terms. LSAs are forward commitments — backlog, not revenue, and none of it has flown. The critical payment-term caveat: these contracts predate a single Terran R orbital flight, so the entire $2.9B is contingent on the vehicle working and arriving on a schedule customers can plan constellations around. Concentration is meaningful — OneWeb alone is ~40%+ of disclosed backlog.

Supply Chain

Upstream inputs → Relativity → end customer, with named stakeholders:

• Raw materials / metal feedstock → Stargate printers. Relativity's vertical-integration story rests on its 4th-gen Stargate large-format metal AM printers (Wire Arc Additive Manufacturing + Powder Bed Fusion) at the ~1M-sq-ft Long Beach plant — the former Boeing C-17 facility, taken over 2022.
• Engines (in-house). Aeon R engines, all active valves, and combustion devices are designed and manufactured in-house — the deepest vertical-integration moat the company has.
• Outside suppliers (the chokepoints). Critically, Relativity uses outside suppliers for fuel-tank domes and fairings — the very large straight-barrel structures it once intended to print but couldn't reliably (a "large buckling event" on a printed dome is documented in internal reporting). This is the single-source / external-dependency risk: the pivot to hybrid manufacturing means Relativity now depends on conventional aerospace suppliers for primary structures it cannot yet print at rate.
• Test infrastructure → NASA Stennis. Engine and stage testing happens at NASA Stennis Space Center, MS, on a 150+-acre expansion with new test stands. This is a dependency on NASA-leased infrastructure and test-stand availability.
• Launch site → Cape Canaveral LC-16. First flights from Launch Complex 16, "rapidly approaching final configuration" as of June 2026.
• End customer: satellite-constellation operators (OneWeb/Eutelsat, SES, Intelsat), NASA, US Space Force, and — prospectively — Schmidt's own orbital-compute ambitions (a captive future buyer).

Chokepoint summary: in-house engines are the strength; externally-sourced tank domes/fairings + a single under-construction launch pad + an unproven stage-test campaign are the fragile links.

Competitive Advantages (moats)

This is the lens where the original story has to be re-underwritten honestly. The "100x part reduction, print-the-whole-rocket" moat largely collapsed. Relativity discovered that printing massive barrel sections "was slowing down production without adding significant engineering value" and pivoted to a hybrid model — in the words of industry coverage, going "from a 3D printing company trying to build a rocket to a launch company that happens to use 3D printing". The dream of a ~95%-printed rocket is dead.

What durable moat actually remains:

• In-house propulsion + AM applied where it pays. 3D printing is now concentrated on engines — the most part-dense, complex, high-value components — which is where AM genuinely has an economic edge. Aeon R is fully in-house with 300+ hot-fire tests and 2,500+ sec of flight-intent firing. This is real, but it is the same kind of moat SpaceX, Rocket Lab, and Blue Origin already have (in-house engines), not a unique one.
• Capital + a singular sponsor. The genuine post-2025 moat is Eric Schmidt's balance sheet and conviction — a ~$34B-net-worth sponsor who has been bankrolling the company since its prior fundraising dried up (Oct 2024) and reportedly committed approaching $800M. That is a moat against the thing that actually kills launch startups (running out of money pre-flight), not a moat against competitors' rockets.
• Backlog as a moat. $2.9B of signed demand before flight signals the market is "desperate for another cost-effective launch services provider" (CRO Josh Brost) — a demand-side pull few rivals have.

Bargaining power: weak today. Pre-flight, Relativity needs its customers more than they need it — every named customer (SES, Intelsat, OneWeb) has alternatives (SpaceX, soon Neutron/New Glenn). Power inverts only if Terran R flies reliably into a supply-constrained market. Over suppliers (dome/fairing vendors) it has modest leverage given low volumes.

Verdict on the moat: narrow and contingent. The defensible asset is in-house propulsion + a uniquely deep-pocketed, strategically-motivated owner — not the 3D-printing mystique that built the brand.

Segments

our figures is empty (header-only) — no research-layer segment data exists. From public sources, the business is effectively single-segment (launch services) with a negligible AM-parts side line. No geographic or product-line P&L breakout is disclosed for a private pre-revenue-scale company.

• Launch services backlog: ~$2.9B, dominated by commercial GEO/LEO constellation operators (OneWeb ~$1.2B, plus SES/Intelsat) with a government slice (NASA VADR; historical DoD ceiling contracts cited up to ~$986M in 2021). These figures are not reconcilable into a clean segment table from public data — flagged as n/a — not disclosed at the line-item level.
• AM parts / Stargate-as-a-service: small, episodic (US Air Force parts) — no disclosed revenue.
• Trend: the only directional read is backlog growth ($1.8B → $2.9B, 2023→2025) — accelerating commitments, driven by industry-wide launch-supply scarcity, not accelerating revenue. The cause is the SpaceX-dependency hedge: constellation operators want a second reliable ride.

Phase B — Measure performance (+private overlay: funding/valuation trajectory · traction · catalysts)

Funding & Valuation Trajectory

No earnings exist. The scoreboard for a private is the round history and burn:

Total raised: sources conflict — Contrary cites ~$1.3–1.6B (2023), later trackers cite "$2.4B+" inclusive of Schmidt. Conflict surfaced, not resolved — pre-Schmidt cumulative ≈ $1.3–1.6B; the ~$800M Schmidt tranche pushes the all-in figure toward ~$2.4B.

Burn signal (the important number): Bloomberg reported Relativity had "burned through more than $1 billion" without realizing the original 3D-print vision and was "running low on cash" in late 2024. The prior fundraise "dried up" around October 2024; Schmidt has been bankrolling it since. Headcount: ~1,200 (2024) → ~1,600 (2026). Revenue: one tracker cites $181.5M for 2024 — treat with heavy skepticism (likely conflates contract bookings/ARR-style figures with recognized revenue for a pre-flight launcher); flag as unverified.

The decisive fact of this lens: the last priced valuation is the June 2021 $4.2B, already marked −21% by a lead investor in 2022, and no new priced valuation has been disclosed under Schmidt. Any "$4.2B" you see quoted in 2026 is a stale 2021 sticker, not a current mark.

Founder / Leadership Interviews (sentiment trend) (calls → founder media)

No earnings calls. Sentiment proxy = leadership's public posture across the 2023→2026 arc:

• 2023 (pivot): defensive but rationalized — retiring Terran 1 and redesigning Terran R framed as responding to "overwhelming market demand". Subtext: the all-printed thesis wasn't working.
• Mar 2025 (transition): Tim Ellis's LinkedIn handoff to Schmidt was graceful — Ellis stays as co-founder/board member. Tone: relief at securing a sponsor.
• 2025–26 (Schmidt era): confident, capability-focused. CFO/President Mo Shahzad: "We have all the elements to make Terran R a commercially competitive launch vehicle." CRO Josh Brost: the market is "desperate for another cost-effective launch services provider." Schmidt himself: the orbital-data-center "Yes".

The shift over time: from evangelizing a manufacturing revolution (2021) → defending a humbling pivot (2023) → executing a focused launch program under a billionaire operator with a much larger AI-infrastructure thesis (2025–26). What they stopped saying: "fully 3D-printed rocket," "95% printed." What they started saying: "commercially competitive," "medium-to-heavy," "data centers in orbit." Sentiment is genuinely more credible now than in 2023 — but it is sponsor-driven credibility, still unbacked by a flight.

Cap Table & Secondary Marks (comps → cap table; plus mechanism/peer comps)

Cap table quality. The syndicate is genuinely tier-1: Fidelity, BlackRock, Coatue, Baillie Gifford, Tiger Global, Soroban, XN, K5, Tribe, Lux, Y Combinator. Crossover funds (Fidelity, BlackRock, Baillie Gifford) are present — historically an IPO-proximity tell — but the most important cap-table fact is the 2025 control change: Schmidt took a controlling stake, which means the venture syndicate's preferred stack was very likely recapitalized / diluted, and the 2021 marks are stale. Fidelity's −21% Series D markdown (2022) is the only disclosed third-party re-mark and it is down. Secondary platforms (Forge, EquityZen) list the name but no clean recent secondary mark is publicly disclosed — n/a.

Peer set (private + public launch comps, capability-based since multiples don't apply):

Multiples: n/a / not applicable (no public comparable P/E or EV/Sales for a pre-revenue private launcher; do not fabricate).

Read: the cap table is high-quality but down-marked and recapped. Among medium-lift challengers, Relativity has the largest disclosed backlog ($2.9B vs Neutron's $1.1B) and the biggest payload, but it is not the best-capitalized relative to ambition (Stoke raised a $1.34B Series D as a smaller-payload vehicle; SpaceX/Blue Origin are already flying). Its differentiation is sponsor + backlog + payload class, not balance-sheet supremacy.

Stock-Price Catalysts (price moves → funding/program milestones)

No public stock; the "tape" is program + funding milestones that re-rated the private mark or sentiment:

• Jun 2021 — Series E / $4.2B: the high-water valuation mark.
• May 2022 — Fidelity −21% markdown: first public down-signal.
• Mar 2023 — Terran 1 flies, fails to orbit: proved the airframe survived Max-Q but missed orbit — a mixed, ultimately bearish event that fed the pivot.
• Apr 2023 — Terran 1 retired, Terran R redesigned: strategic reset; bearish on the original thesis, bullish on commercial focus.
• Late 2024 — cash crunch / "$1B burned": the existential low.
• Mar 2025 — Schmidt takes control (~$800M): the single biggest positive re-rate of sentiment; converted a near-death story into a credibly-funded one.
• May 2025 — orbital-data-center "Yes": reframed the entire investment narrative from launch pure-play to AI-infrastructure option.
• Nov 2025 — SES expands LSA: commercial validation pre-flight.
• Jun 2026 — second stage to Stennis, LC-16 nearing final config: execution proof points keeping the late-2026 maiden flight credible.

Pattern: this name re-rates on (1) funding/sponsor events and (2) binary program milestones (a flight, a pivot). The dominant future catalyst is unambiguous: the Terran R maiden flight. Everything else is noise around it.

Phase C — Judge people & books

Management

• Eric Schmidt (CEO & Chairman, since Mar 2025). Track record: ran Google as CEO 2001–2011 and was Alphabet executive chairman to 2017 — i.e., scaled one of history's great cash-flow machines, but as a professional operator/steward, not a hardware founder. Skin in the game: maximal and unusual — he didn't just invest, he took control with ~$800M of his own ~$34B net worth and is personally bankrolling the company. Capital-allocation history: at Google, strong; in deep-tech hardware and space, unproven. Archetype: deep-pocketed strategic owner-operator, not a rocket engineer — which means execution risk on a hard physics problem sits with the team beneath him.
• Tim Ellis (co-founder, board). Built Relativity from nothing to a $4.2B mark and an orbital launch attempt — genuine founder credibility — but also presided over the $1B+ burn and the strategy that had to be reset. Stepped back gracefully; remains on the board.
• Jordan Noone (co-founder). Stepped from CTO to advisory in 2020.
• Mo Shahzad (President/CFO), Josh Brost (CRO): the operating bench carrying commercial + finance.

Red flags (management): (1) the control change itself — a founder-led venture became a single-billionaire-controlled company; minority/employee economics in that recap are undisclosed and could be unfavorable; (2) a dual agenda — Schmidt's orbital-data-center thesis may not be aligned with near-term launch-customer interests (is Terran R being built to serve SES, or to serve Schmidt's compute ambitions?); (3) the historical strategy pivot (all-printed → hybrid) is a real "we were wrong" admission, to their credit handled, but it cost years and >$1B. Founder vs. professional: this is now a professional-operator-controlled company with founders relegated to the board — appropriate for a capital-intensive scale-up phase, but it means the visionary risk-tolerance that built it is no longer in the driver's seat.

Forensic Red Flags

No audited financials exist — Relativity is private with no CIK and files nothing with the SEC. There is no income statement, balance sheet, or cash-flow statement to forensically dissect; every standard red-flag test (revenue recognition, receivables vs. revenue, SBC, goodwill) is n/a — private, not disclosed. The honest forensic posture for a private pre-flight launcher is:

• Backlog ≠ revenue. The most important "accounting" caveat: the headline $2.9B is signed contracts contingent on a vehicle that has never flown. It is not recognized revenue and may include termination/renegotiation rights that don't survive a slip or failure. Treat backlog as demand signal, not money.
• The $181.5M "2024 revenue" figure is unverified and likely conflates bookings/ARR-style metrics — do not rely on it.
• Burn opacity: >$1B burned through 2024 is the only hard cash datapoint; current runwayRunwayHow long the cash lasts at the current rate of spending. It shortens the moment spending rises, which is why a figure taken from a quiet quarter flatters. depends entirely on undisclosed Schmidt funding terms.

Regulatory findings (required). Per the pre-fetched regulatory/regulatory-findings.md (generated 2026-06-18, sources SEC EDGAR EFTS LR + AAER): 0 SEC findings — and none are possible, because Relativity has no CIK and is not an SEC registrant. Non-SEC web check ("Relativity Space" (FTC OR DOJ OR FDA OR CFPB OR consent decree OR settlement OR fine OR penalty) enforcement): no material enforcement actions, consent decrees, fines, or penalties surfaced in public search as of 2026-06-18. There is no 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. Item 3 to quote (private). Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (no CIK / not a registrant), web search, and the absence of any public-company disclosure, as of 2026-06-18. (Caveat: a private company's litigation is not publicly indexed the way a registrant's is, so absence of evidence is weaker here than for a public name.)

Phase D — Project & stress-test

IPO-Readiness & Path-to-Tradeable (forward EPS → IPO readiness)

No EPS to project — the question for a private is what makes it tradeable, and when. Using the private-watch.json readiness scale (1 early → 5 S-1 filed/imminent):

• Current stage: late-stage / pre-IPO, but control-recapped — readiness ≈ 3. It has tier-1 + crossover backers and a $2.9B backlog (IPO-narrative ingredients), but the gating fact is that it is a single-billionaire-controlled, pre-first-flight company whose last priced round was 2021 and got marked down. A controlled company with a sponsor willing to fund it privately has less pressure to IPO, not more.
• Milestones that unlock an S-1 (the real catalyst ladder):

1. Terran R maiden flight (late 2026) — the binary that gates everything. No credible IPO before first orbit.
2. Demonstrated reuse + cadence (likely 2027–28) — proving the unit economics that justify a public multiple.
3. Backlog conversion to recognized revenue — turning the $2.9B into flown, recognized launches.
4. A clarified data-center strategy — markets will want to know if they're buying a launch company or an orbital-compute company.

• Estimated IPO window: not before 2028, plausibly later — and possibly never as an independent IPO if Schmidt prefers to keep it private/captive to a broader AI-infrastructure play, or fold it into a larger vehicle.
• Path-to-tradeable today: secondary/SPV only (Forge, EquityZen, the MCXGP SPV) at undisclosed/stale marks — illiquid and opaque.

No Brier forecast logged (--watchlist mode; our model skipped). The natural binary to track if promoted: "Terran R reaches orbit on its maiden flight before 2027-06-30" — base-rate for a maiden orbital flight of a new medium/heavy vehicle is low, ~30–40%.

Write-back note: Relativity is absent from research/private-watch.json. Per strict wave boundaries I am not editing that file in this run, but it should be added (suggested: beat: space, stage: late-stage, ipo_readiness: 3, lead_investors: "Eric Schmidt (control), Fidelity, BlackRock, Coatue", catalyst: "Terran R maiden flight late-2026; orbital-data-center optionality", dossier: this file) in a follow-up conversational pass.

Bull vs Bear

Bull case. A genuine medium-to-heavy reusable gap exists between Electron-class small lift and Falcon 9, and demand is real ($2.9B signed, SES expanding in 2025, NSSL Phase 3 + SDA + mega-constellations all needing a second ride). Relativity has the largest payload and largest backlog of the medium-lift challengers, in-house propulsion with thousands of seconds of test time, and — the differentiator — Eric Schmidt's balance sheet and a much larger thesis (orbital AI data centers) that could make Terran R the launch backbone of a category Schmidt is uniquely positioned to will into existence. If Terran R flies on schedule into a SpaceX-dependent, supply-starved market, the bargaining power inverts and the $4.2B mark looks cheap.

Bear case (permanent-impairment risks). (1) It has never reached orbit. The one orbital attempt (Terran 1) failed on the second stage, and the company has since reset its entire vehicle and manufacturing strategy — Terran R is effectively a clean-sheet rocket that has not flown. New medium/heavy launchers slip and fail routinely (Neutron already slipped to 2026; New Glenn took years). (2) The competitive window is closing fast — by the time Terran R debuts (late 2026 if on schedule), it lands into Falcon 9 (operational, 165 flights/yr), New Glenn (operational, ~45t, larger), and Neutron (13t, flying ~2026). It will be the last of the medium-lift class to fly. (3) The valuation is fiction-until-proven — the only priced mark is a stale, down-marked 2021 $4.2B; a control recap likely impaired prior holders, and there is no current priced round.

Pre-mortem (18 months out, thesis broke): Terran R's maiden flight slipped to 2027 or failed on ascent; meanwhile New Glenn and Neutron captured the medium-heavy commercial manifest; one or two anchor customers (an SES or OneWeb) quietly re-papered to a flying provider; and Schmidt's orbital-data-center thesis proved thermally/economically unworkable in the near term, removing the captive-demand call. The mark gets cut hard in the next financing.

Are multiples too high? No public multiple exists, but the implied $4.2B-anchored expectation is too high for a pre-flight launcher entering last —. A clean current mark would likely be flat-to-down vs. 2021 absent flight proof.

Contrarian view (what the market refuses to see): The crowd narrative is "Schmidt + AI data centers in orbit = moonshot upside." The contrarian read is the opposite — Schmidt's involvement is a rescue, not a rocket booster. He's bankrolling a company that burned $1B and couldn't raise, and the orbital-data-center story may be doing more work supporting the valuation than the (still unflown, last-to-market) rocket can justify. The single thing that matters — does Terran R reach orbit on schedule — is being underweighted relative to the shiny AI framing.

Devil's Advocate (short-seller)

Dismantling the bull case as a skeptic:

• What structurally breaks the model: Relativity makes money only if Terran R flies reliably and cheaply and arrives before the medium-heavy market is locked up. It is last to the party — New Glenn and Falcon 9 are already operational, Neutron is close. A launch company with no flown orbital vehicle and a clean-sheet design has not de-risked the only thing that matters.
• Revenue concentration: OneWeb is ~40%+ of disclosed backlog. OneWeb/Eutelsat's own constellation economics are shaky; if that customer slips, re-scopes, or defects to a flying provider (as Iridium largely did to SpaceX), a huge chunk of the "backlog moat" evaporates. Backlog signed pre-flight is the softest kind.
• Why the moat is weaker than bulls think: the 3D-printing moat was abandoned — they now buy domes and fairings from conventional suppliers. In-house engines are table stakes, not a differentiator. The real "moat" is one billionaire's checkbook, which is a dependency, not a durable competitive advantage.
• Most dangerous competitor bulls underestimate: Blue Origin's New Glenn — already flying, larger payload, Bezos-funded, and chasing the same heavy commercial + national-security manifest. And SpaceX, which can simply drop Falcon 9 pricing to defend share. Bulls fixate on the medium-lift "gap" that New Glenn + Neutron are about to fill first.
• Worst capital-allocation / governance moves: $1B+ burned on a manufacturing thesis that was then largely scrapped; a control recapitalization with undisclosed terms for minority/employee holders; and a CEO whose primary strategic interest (orbital data centers) may not be aligned with paying launch customers.
• Assumptions that must hold for today's price: Terran R flies on schedule (late 2026), achieves reuse, converts $2.9B backlog to flown revenue, and the orbital-data-center call provides incremental captive demand. If growth/timeline disappoints 20–30% (a one-to-two-year slip + a customer defection), a clean valuation is plausibly well below the 2021 $4.2B.
• Single scenario that permanently impairs the business: a maiden-flight failure (the Terran 1 precedent) combined with New Glenn/Neutron capturing the manifest — leaving Relativity to re-fly and re-prove while customers have moved on. Plausibility: moderate — maiden orbital flights of clean-sheet vehicles fail or slip more often than not.

Management Questions (ordered by information value)

1. What are the precise terms of Eric Schmidt's controlling investment — equity stake, preference, and the dilution/treatment of prior venture holders and employee option holders in the recap?
2. What is the current cash runway (in months) and does it fully fund Terran R through a successful maiden flight, including a likely first-flight failure and refly? At what burn rate?
3. Is Terran R being optimized for current paying customers (SES/Intelsat/OneWeb) or for the orbital-data-center thesis — and where do those requirements conflict?
4. What is the realistic, internally-held probability and date for the Terran R maiden flight reaching orbit — and what is the refly plan and timeline if it fails?
5. Of the $2.9B backlog, how much carries firm, non-cancellable commitments that survive a 12-month schedule slip, and what is the largest single-customer concentration?
6. What is the credible per-launch cost (not price) at first commercial cadence, and the price/kg at which Terran R is unit-economics positive vs. Falcon 9 and New Glenn?
7. The orbital-data-center vision faces unsolved thermal-rejection, power, and servicing physics — what is the concrete technical and capital roadmap, and what does Relativity build vs. partner for?
8. With New Glenn operational and Neutron near, why does the market need Terran R specifically — what is defensible once three reusable medium-heavy vehicles are flying?
9. Which structures are still externally sourced (domes, fairings) and what is the supplier-concentration / single-source risk on those primary structures?
10. What is the path and timeline to NSSL Phase 3 / national-security certification, and how much of the future TAM depends on it? (Note: an NSSL bid was reportedly delayed to focus on the 2026 debut.)
11. Aeon R has 2,500+ sec of flight-intent firing — what is the demonstrated reliability margin and the full-duration, full-stage qualification status as of today?
12. What is the realistic Terran R flight cadence in years 1–3 post-debut, and what gates the ramp (engines, pads, stages)?
13. What is the company's preferred liquidity path — independent IPO, strategic combination with a larger Schmidt AI-infrastructure vehicle, or staying private — and on what timeline?
14. What did the Terran 1 second-stage failure teach you that is specifically designed out of Terran R's second stage and upper-stage propulsion?
15. What is the single technical or commercial risk that most keeps management up at night, and what is the mitigation?