The Lens · Deep Dive

The SpaceX IPO, Explained

The biggest stock-market debut in history just happened — SpaceX priced at $135 and jumped ~19% to about $161 on day one, a roughly $2.1 trillion company. Here's the whole thing — in plain English, from scratch — including the Tesla convergence, the chip fab, the AI software project most people have never heard of, and a map of exactly who profits and who's in trouble. Hover (or tap) any underlined term for a definition.

Dragonfly Lens · Updated June 15, 2026 (post-IPO) · ~25 min read · Built to be a reference, not a hot take.

The 60-second version

What's inside 1. What SpaceX actually is 2. The numbers 3. How the IPO works 4. Is it worth $1.75T? 5. The Tesla convergence 6. Terafab & Macrohard 7. Bull / base / bear 8. Who profits, who's at risk 9. The tech-IPO report card 10. How to think about entry Glossary

1. What SpaceX actually is (it's not just rockets)

Most people picture SpaceX as a rocket company. That's the smallest part of the story now. It's really five businesses stacked together:

The full anatomy → Since the Feb 2026 xAI merger, SpaceX also rents AI compute (Anthropic pays $1.25B/mo, Google $920M/mo). For the complete picture — who owns it (Musk: 42% equity, 82% votes), every revenue stream, the full board of variables that move the stock, and the honest "Elon effect" — see SpaceX: The Complete Picture →

So when you buy SPCX, you're not buying a rocket company. You're buying a conglomerate — with one foot in proven, profitable infrastructure and the other in expensive, unproven frontier bets.

2. The numbers (in plain English)

$18.7B
2025 revenue (+33% vs 2024)
−$4.9B
2025 net loss (mostly from xAI)
$11.4B
Starlink revenue (61% of total)
$4.4B
Starlink operating profit (the only profitable unit)

Read those four boxes together and the story is clear: Starlink makes real money, and almost everything else spends it. The company brought in $18.7 billion but still lost ~$4.9 billion overall — because the newly-absorbed AI business (xAI) burned roughly $6.4 billion on its own, and SpaceX spent about $20.7 billion on capex building rockets, satellites and AI infrastructure. from the S-1

Watch the trend underneath the growth: Starlink's revenue per customer is falling — from about $99/month (2023) to ~$66/month (early 2026). Subscribers are up huge, but the company is partly buying that growth with lower prices.
The revenue line the 2025 numbers don't show yet: AI-compute rentals. Since the xAI merger, SpaceX rents data-center compute to its own AI rivals — Anthropic pays $1.25B/month (through May 2029, ~$40B+) and Google pays $920M/month (confirmed Jun 2026: ~110,000 Nvidia GPUs, Oct 2026–Jun 2029, ~$30B total). Together that's a ~$26B/year run-ratemore than all of SpaceX's 2025 revenue combined — ramping from late 2026. The honest caveats: it's concentrated in just two tenants, both deals carry 90-day cancellation options, and the Google leg has a GPU-delivery milestone (Sep 30, 2026). Real, enormous, and not guaranteed — the full breakdown is in The Complete Picture. confirmed

Source: SpaceX S-1 filing (filed May 20, 2026) and reporting on it — PitchBook, MostlyMetrics.

3. How the IPO actually works

ThingDetail
Ticker / exchangeSPCX on the Nasdaq
First tradeJune 12, 2026 — closed day one near $161 (+19%) from the $135 IPO price confirmed
Price$135/share — a fixed price, not a range; ~$2.1T market cap after the day-one pop confirmed
Raise~$75 billion (~556M shares)
Where the money goesAll-primary — all proceeds go to SpaceX; no insiders selling confirmed
ControlDual-class — Musk keeps 82–85% of votes with ~42% of equity confirmed
Retail accessDirect via Robinhood, SoFi (no minimum), Fidelity ($2,000), Schwab ($100k). Allocations limited.
LockupLockupstaggered, not one 180-day cliff. ~20% unlocks after Q2 earnings, more in tranches (~70/90/105/120/135 days), rest by ~180 days. Musk stays fully locked. confirmed

Two things to internalize. First, because it's all-primary, this raise is SpaceX fueling itself — not insiders cashing out (a healthier signal). Second, the dual-class structure means SpaceX is a “Controlled Company”: you're a passenger, not a co-pilot. If you don't trust Musk's capital allocation, there is no governance lever to pull.

The lockup myth, corrected. A viral claim says SpaceX insiders “can't sell for 368 days, and only 30% above the IPO price.” That's wrong, and backwards. The filing's lockup is staggered to let most insiders sell earlier than a normal IPO, not later: ~20% unlocks after the first quarterly earnings as a public company; an extra slice frees up early if the stock holds 30%+ above the IPO price for five of ten days (a bonus accelerator, not a price floor); then time-based tranches at ~70, 90, 105, 120, 135 days, with the remainder by ~180 days. There is no “368 days.” The one real “can't sell” is Elon Musk, carved out and kept under the full restriction. Why it matters: staggered supply means insider selling starts sooner but arrives smoother — it softens the classic single-day 180-day cliff that hammers most IPOs (see the report card in §9).
Going deeper → what actually happens AFTER it lists. The listing is day 1 — but the moves that matter come after: forced Nasdaq-100 index buying (~day 15), a tiny-float squeeze, why the S&P 500 can't buy in yet, and the lockup supply flood ~3–6 months out. The full supply-and-demand timeline of the biggest IPO ever: The SpaceX IPO Aftermath — the mechanics nobody's pricing →

4. Is it actually worth $1.75 trillion?

This is the question, and the honest answer is: nobody agrees.

$1.75T
IPO target valuation
~$780B
Morningstar's fair-value estimate (−48%)
~94×
Price-to-revenue at the target (extremely high)

At $1.75 trillion you'd be paying about 94 times annual revenue — a multiple normally reserved for tiny hyper-growth software, not a capital-heavy company that lost ~$5 billion last year. The independent research firm Morningstar publicly values SpaceX near $780 billion — less than half the asking price — and called the share structure a long list of “unfriendly” terms.

The reference-grade takeaway: the $1.75T figure is a seller's target during peak hype, not an agreed value. Treat any single valuation number — including the bullish ones — as a claim to be checked, not a fact.
How wildly the dream is priced — read Musk's own pay package. SpaceX's 2026 compensation plan tells you exactly what investors are underwriting. The top tranche of Musk's award only vests if SpaceX reaches a $7.5 trillion valuation and a permanent 1-million-person Mars colony; a separate target needs SpaceX operating 100+ terawatts of data-center power in space — over 1,000× every data center on Earth combined today. Miss the targets and he collects a $54,080 salary (his pay since 2019). The board tied his comp to the mission itself, which is inspiring — and also a flashing sign that at ~$2.1T you are paying today for outcomes that are decades away and entirely unproven. confirmed

5. The Tesla convergence (why this is really about Musk's whole empire)

You can't understand SpaceX in 2026 without Tesla, because Musk is quietly welding his companies into one machine. Here's Tesla by the numbers first, because half the SpaceX story now runs through it.

Tesla, briefly (most recent quarter, Q1 2026)

MetricQ1 2026What it tells you
Revenue$22.4B (+16% YoY)Still growing, but missed expectations
Gross margin21.1%Improved — but partly on one-time warranty/tariff items
Net income$477MThin for a company valued in the trillions
Cash$44.7BFortress balance sheet — the real strength
Vehicle deliveries358k (−14% vs Q4)FY2025 fell 8.6% — 2nd straight down year
Energy storage8.8 GWhFastest-growing segment, but lumpy

The Tesla debate in one line: the car business is shrinking, so the trillion-dollar valuation rests entirely on the “optionality — robotaxis, the Optimus humanoid robot, in-house AI chips, and energy. As of mid-2026 the robotaxi fleet is ~20–39 cars in Austin and Optimus revenue is roughly zero. The bull case is “these become enormous.” The bear case is “you're paying a ~350× P/E for a car company selling fewer cars.”

How the companies are folding together confirmed via the S-1

On the “Tesla–SpaceX merger”: there is no official merger. Reporting frames a combined ~$3.4T “empire,” and betting markets put a formal merger at only ~5% by mid-2026 but ~50% by 2027. The convergence is already real operationally (shared ownership, purchases, projects); the legal merger is speculation. speculative

6. Terafab & Macrohard — the parts nobody knows

Terafab — the chip factory

Announced March 21, 2026, Terafab is a giant chip fab being built near Austin, Texas by Tesla + SpaceX, with Intel joining in April 2026 as the manufacturing partner (using Intel's cutting-edge 14A process). The pitch: make Tesla's AI5 chips (for self-driving and the Optimus robot) and special “non-terrestrial” chips for AI data centers in space. The stated goal is absurdly large: one terawatt of computing power per year.

Cost estimates range wildly: a $25B headline at announcement → SpaceX's own filing says $55B to start and up to $119B for all phases.

The bear view, which you should hear: tech outlet Electrek called Terafab a project that “reeks of desperation” — arguing it conveniently attaches Tesla's declining car business to SpaceX's hot IPO. Their math: 1 million wafers/month would be ~70% of TSMC's entire global output — from companies that have never built a chip. And crucially…
…the S-1 admits it isn't real yet. Per SpaceX's own filing, Terafab (and Macrohard) are in “very early stages” with no financial terms, no IP rights, and no binding commitments — “neither Tesla nor Intel are obligated to remain.” Treat Terafab as a vision, not a factory. non-binding per S-1

Macrohard — the AI that wants to clone software companies

Macrohard (a deliberate joke on “Microsoft”) is a Tesla–xAI project to build software entirely from cooperating AI agents. The idea: point Grok (as the “brain”) plus a Tesla-built agent that watches a screen and controls a mouse/keyboard, and have it replicate what entire software companies do. Musk's framing is a direct shot at Microsoft.

Reality check: the trademark was filed Aug 2025; the project was unveiled March 2026. There is no product, no revenue, and no customers — it's a stated capability with no public demonstration. The technology (AI that reliably operates software like a human) is genuinely unproven across the whole industry. vision-stage

Why these matter anyway: the vertical stack

Even if Terafab and Macrohard are years early, they reveal the strategy — a single, self-contained loop Musk is trying to own end to end:

Chips (Terafab) → AI models (xAI / Grok) → AI software (Macrohard) → robots & cars (Tesla / Optimus) → energy (Megapack) → space & internet (SpaceX / Starlink) → back to chips

If even half of it works, it threatens Microsoft (software), TSMC (chips), Nvidia (AI hardware) and the big cloud providers at once. That ambition is exactly why the IPO commands attention — and exactly why the risk is enormous.

7. The trajectory: bull, base, bear

▲ Bull case

Starlink keeps compounding (tens of thousands more satellites, global demand); defense becomes critical national infrastructure (the $2.29BGolden Dome” deal, with the program ceiling raised to $13B); the launch monopoly widens with Starship; and the AI/orbital-compute bet becomes the trillion-dollar option on top.

▬ Base case

A great infrastructure company (internet + launch + defense) growing 25–35%/yr, eventually profitable as xAI losses are reined in — worth a lot, but not obviously $1.75T on day one. The AI moonshots stay optional.

▼ Bear case

Key-man risk (Musk's 82–85% control, his politics flagged as a business risk in the filing); relentless cash burn ($20.7B capex, AI losing billions); real competition (Amazon's Kuiper/Leo, AST SpaceMobile); and regulation (FAA launch limits, FCC spectrum fights, and orbital-debris backlash). At 94× revenue, a lot has to go right.

8. The ripple effect: who profits, who's at risk

This is where it gets investable. A company this big doesn't rise alone — it drags whole supply chains up and pushes incumbents down. Here's the map. Important: the cleanest pure-plays are often private (no ticker you can buy); we flag those clearly.

▲ Upstream — who builds the picks and shovels

NameTickerWhy it benefits
IntelINTCThe single cleanest confirmed winner — named Terafab foundry partner. confirmed
Howmet AerospaceHWMTitanium & superalloy launch structures, fasteners
HEICOHEISatellite thermal/electronic components — scales with sat count
Karman, TransDigm, LoarKRMN, TDG, LOAREngineered aerospace/defense parts inferred
Advanced Mfg. Co. of AmericaPRIVATEEx-SpaceX-led rollup of US defense manufacturers (sensors, hydraulics, avionics). Not tradeable.

▲ Second-order — the power to run it all

Terafab's “one terawatt of compute” and the broader AI build need staggering amounts of electricity. Texas grid demand could quadruple by 2032. That spotlights uranium & nuclear: miners CCJ, UEC, UUUU, NXE, enricher LEU, and reactor builders OKLO, SMR, NNE. (These ride the whole AI-power wave; Terafab is additive demand, not a signed contract.) indirect

▲ Space tracking — the traffic-control layer

More satellites = more near-misses. Starlink alone ran ~300,000 collision-avoidance maneuvers in 2025 (+50% YoY). That makes space-traffic tracking mandatory. Public: Rocket Lab (RKLB) — real $816M + $90M defense/tracking contracts; Redwire (RDW) — record backlog. Private pure-plays: LeoLabs (radar) and Slingshot (optical) — the best expressions, but not tradeable.

▼ Downstream — who's in trouble

At riskTickerWhy
GEO satellite operatorsVSAT, SATSHighest-conviction losers. One industry peer said its GEO consumer value is now “zero.” EchoStar already sold $17B of spectrum to Starlink.
Phone carriersT, VZ, TMUSForced into an unprecedented joint venture (May 2026) to pool spectrum and defend against Starlink direct-to-cell.
Legacy launch(private)ULA, Arianespace, Blue Origin — out-priced by reusability
Space insurersPremiums reportedly up ~400%; some crowded orbits heading toward “uninsurable” speculative

The insurance angle — a market in retreat (and why it matters)

Here's the counterintuitive part most coverage misses: as orbits get crowded, insurers are running away from the risk, not toward it. Major underwriters — Swiss Re, Allianz, AIG, Brit, Aspen Re — have exited space insurance entirely, and the Lloyd's specialist Canopius stopped writing it. The whole global space-insurance market is tiny (~$500–900M/year in premiums) and just lived through its worst loss years on record (~$1B in 2023 claims, including a record $420M Viasat payout).

Why there's no “space-insurance stock” to buy: space is underwritten by a handful of Lloyd's of London syndicates and is an immaterial slice of giant diversified (re)insurers (Munich Re, Swiss Re, AXA) and brokers (AON, Marsh / MMC). There is no public pure-play — and the names with exposure are cutting it, not growing it.

So the real investable signal is second-order: debris premiums for affected orbits are projected up ~400%, and some orbital shells are heading toward literally uninsurable (a state actor can “de-license” an altitude by seeding it with debris). When you can't insure a satellite, your only defense is to track and dodge it — which is exactly why the insurance retreat reinforces the space-tracking (SSA) thesis above (LeoLabs, Slingshot, RKLB). A nascent “space-debris collision insurance” niche is just emerging (first-of-kind products from specialist startups — private). reported

The synthesis: the single cleanest confirmed public beneficiary is Intel (the Terafab partner). The cleanest at-risk cohort is legacy GEO satellite operators (VSAT, SATS) and, more slowly, the phone carriers. The highest-leverage pure-plays (AMCA, LeoLabs, Slingshot) are private — which is part of why a SpaceX listing matters: it cracks the window open for the whole private space complex to follow.

9. The tech-IPO report card (history rhymes)

The single most useful thing before any hot IPO is to look at what happened to the last batch. The pattern is remarkably consistent: a day-one pop, months of euphoria, then a brutal drop — and, for the survivors, enormous long-run gains if you bought the wreckage, not the hype.

CompanyIPOThen → worst dropWhere it stands today
Amazon1997−96% (dot-com bust)Up ~1,800× from IPO — survivors compound
Facebook (Meta)2012−54% in monthsUp >10× — the post-crash buyers won
Nvidia1999−90% (multiple times)Among the most valuable firms on Earth
Coinbase2021−92%Recovered hard — but only for those who held/added
Snowflake2020−70%Still below its hype peak years later
Rivian2021−85%+Cautionary tale — hype > fundamentals
Lyft2019−85%Never recovered its IPO price
CoreWeave2025−55%The most recent AI-IPO round-trip

Across recent high-profile tech IPOs, the average drawdown from the early peak was roughly −74%, bottoming around 10 months after listing — which, not coincidentally, is right around when lockups expire and insider selling peaks. The lesson isn't “avoid” — it's that day-one is almost never the entry, and the business quality (Amazon vs Lyft) decides whether the dip is an opportunity or a trap.

10. How to think about entry (not advice — a framework)

Our read, in one paragraph

SpaceX is a genuinely historic company opening a genuinely new frontier — and the IPO is being priced for the dream, not the proof. The cash engine (Starlink + defense) is real and excellent; the AI/chip/orbital stack is a high-variance option the company itself calls unproven; and the price (~$1.75T) is contested by half. If history rhymes, the patient story beats the day-one story. We're watching it as the catalyst that re-rates the entire space complex — the supply chain, the tracking layer, and the dozens of private names waiting in line behind it.

Glossary (everything in one place)

IPO — Initial Public Offering: the first time a private company sells shares to the public on a stock exchange.

S-1 — the detailed disclosure a company files with the SEC before its IPO; the most honest document about its risks.

Valuation — what the whole company is judged to be worth (share price × number of shares).

Dual-class shares — insiders hold extra-vote shares so a founder keeps control with a minority of the economics.

Lockup — the period after an IPO when insiders can't sell. Usually one 180-day cliff; SpaceX's is an unusual staggered release (waves starting ~70 days, ~20% after Q2 earnings), with Musk kept fully locked.

Capex — money spent building long-lived assets (factories, satellites); heavy capex burns cash now for future capacity.

ARPU — average revenue per user; falling ARPU means growth is coming partly from price cuts.

P/E ratio — price divided by per-share earnings; high P/E = paying a lot for current profit on a growth bet.

Starlink — SpaceX's satellite-internet constellation; its profit engine.

xAI / Grok — Musk's AI company and its chatbot, now owned by SpaceX.

Terafab — the proposed Tesla/SpaceX/Intel chip factory (very early stage).

Macrohard — a Tesla–xAI project to build software from AI agents (vision stage, no product).

GEO vs LEO — geostationary orbit (high, slow, old) vs low-Earth orbit (low, fast, Starlink). LEO is displacing GEO for consumer internet.

SSA — Space Situational Awareness: tracking orbital objects to avoid collisions.

Kessler syndrome — a runaway chain of satellite collisions that could make an orbit unusable.

If this was useful

This is what we do every week.

Dragonfly Lens breaks down a frontier company or market shift like this one — plain English, every claim sourced and flagged, no hype. Get the next one before the crowd does.

Join the Lens →
More from the Lens: The AI IPO Wave tracker · All explainers · Dragonfly Lens home

Sources: IPO price / date / $1.75T valuation — CNBC, Bloomberg, CBS News; ownership & dual-class (Musk ~42% equity / 82.4% votes) — MEXC, KeepTrack; xAI merger — CBS News; revenue ~$27–30B + Starlink reliance — CNBC, NextBigFuture; Terafab / Intel — EE Times, TechCrunch; defense contracts — SpaceNews, Fed-Spend.

Educational research, not personalized investment advice. Dragonfly Lens is not a registered investment advisor. Every figure here is drawn from public filings and reporting as of June 15, 2026 and is labeled confirmed, reported, or speculative — verify against primary sources (the SpaceX S-1 / 424B prospectus, SEC EDGAR, and company filings) before acting. Pricing, dates, and the valuation are targets that can change at pricing. Past performance does not guarantee future results. We may discuss companies we are watching; nothing here is a recommendation to buy or sell.

Primary sources: SpaceX S-1 (SEC EDGAR, filed May 20 2026) · CNBC · Morningstar · PitchBook · Electrek (Terafab/Macrohard) · SpaceNews · CNBC (Tesla Q1 2026).