SpaceX isn’t coming public as just a rocket company. It’s coming public as a rocket company, a global internet provider, an AI lab, and a social platform — all wrapped into one ticker, all controlled by one person holding 85% of the votes. Starlink is the real engine here, and it’s impressive. But a $1.75 trillion valuation, nearly $5 billion in yearly losses, and a major contract that can walk away with 90 days’ notice make this the most complicated IPO of this generation.
Nasdaq: SPCX · IPO Target June 18–30, 2026
$1.75T–$2T
Roadshow: June 8 · Raise target: $75B (largest IPO in history)
Underwriters: Morgan Stanley · Goldman Sachs · JPMorgan · BofA · Citi
Retail allocation: 30% of float · 5-for-1 stock split already completed
Musk voting control
85%
Class B shares — you have no say
42% economic interest
what you're actually buying — 4 businesses in 1 ticker
| Business |
Revenue (FY2025) |
Profit/Loss |
Simple description |
| Starlink (Connectivity) | $11.4B (+50%) | $4.4B profit | Satellite internet. The only profitable part. 10.3M subscribers. |
| Space (Rockets) | $4.1B | Loss | Falcon 9 launches for NASA, DoD, commercial clients. |
| xAI / Grok / Colossus | Growing | -$6.4B loss | AI chatbot + data centre. Burning $2.5B every 3 months. |
| X (formerly Twitter) | Declining | Loss | Social media. Valued at $33B. Advertiser base has shrunk. |
Plain English summary: Starlink makes money and subsidises everything else. xAI, which includes Grok and a supercomputer in Memphis, is burning through Starlink's profits. Before the xAI merger, SpaceX was earning ~$8B profit per year. After the merger, it is reporting a $4.94B annual loss. You are buying one excellent business and three loss-making ones.
starlink — the reason anyone is buying
2025 Revenue
$11.4B
+49.8% year on year
2025 EBITDA
$7.2B
+86% year on year
Subscribers
10.3M
Crossed 10M Feb 2026
Monthly ARPU
$81
↓18% — growing by lowering price
Satellites in orbit
9,600+
Biggest constellation ever
Enterprise pricing
$3K/mo
vs. $120–250 consumer
overall financials — the full picture
Starlink EBITDA (cash earned)
xAI losses (cash burned)
Total GAAP net result
Starlink earns billions. xAI burns billions. The GAAP "net" is the leftover — a growing loss as xAI scales up.
the anthropic deal — stressed
⚠ Critical — Read This Carefully
$1.25B/month with Anthropic. Not locked in.
SpaceX's Colossus 1 data centre in Memphis has a deal with Anthropic — the AI company behind Claude — worth
$1.25 billion per month, running until May 2029. That sounds like $45 billion in guaranteed revenue. It is not.
Material Uncertainty.
Elon Musk posted on X on May 28, 2026, directly contradicting the S-1's framing of the Anthropic deal. Musk wrote. "This is a 180 day lease with 90 day notice mutual cancellation thereafter. The short term was our request, not Anthropic's." Musk's statement directly contradicts SpaceX's S-1 filing, which presents the deal as running "through May 2029." The disclosure gap — the S-1 says three years, the CEO says six months???
The termination clause
Either party can exit with 90 days' notice — no penalty. There is no lock-in. The $45B total is a maximum possible value, not a guaranteed figure.
Anthropic is majority-funded by Amazon, which operates its own competing cloud (AWS). If Amazon steers Anthropic's compute to AWS, this contract evaporates.
Anthropic can also negotiate pricing down aggressively as more data centre supply comes online (Meta, Microsoft, Google all expanding). Renewal is not guaranteed.
Why it still matters
$1.25B/month is $15B/year if sustained — comparable to Starlink's 2025 revenue. Even 2 years of this contract ($30B) would materially change SpaceX's financial profile.
It proves there is real enterprise demand for SpaceX's AI compute infrastructure — not just a pitch deck. The contract exists today.
But at a $1.75T IPO valuation, the market appears to be pricing this as if it's guaranteed. The 90-day exit clause means it isn't.
the 5 thresholds — what has to go right
01
Power grid bottleneck
Colossus needs gigawatts of electricity. The Memphis power grid wasn't built for this. Expanding grid capacity takes 3–5 years. You can't rush the electricity authority. Production target of 1 million GPUs requires power supply that does not yet exist at that scale.
02
Funding limitation
Capex jumped from $5.6B (2024) to $20.7B (2025) — a near-fivefold increase. Total liabilities now exceed $60.5B. The IPO's $75B raise is partly needed to sustain this spending rate. If AI investment returns don't materialise fast enough, another capital raise is likely within 2 years.
03
Supply chain & engineering lead times
NVIDIA GPU allocations are rationed globally. Cooling systems, power electronics, and data centre components have 12–18 month lead times. SpaceX is not the only company competing for this supply — Meta, Microsoft, and Google are all ordering at scale simultaneously.
04
Talent — 9 of 11 xAI co-founders are gone
Since the merger completed in February 2026, nine of xAI's eleven original co-founders have departed. These were researchers from DeepMind, OpenAI, and Google Brain. The xAI CFO also left in April 2026. Rebuilding this team while integrating three companies simultaneously is a serious operational risk.
05
Orbital AI compute — the "big vision" isn't until 2028 at the earliest
SpaceX's long-term thesis involves putting AI data centres in space — satellites that process AI workloads in orbit using solar power. Timeline: demos late 2026, customer pilots 2027, volume deployment 2028 best case. Investors buying at IPO are funding a concept that won't generate revenue for at minimum 2 full years. This assumes Starship — which has never flown a commercial mission — also succeeds on schedule.
the capital loop — how the banks are exiting through you
The Underwriting Syndicate's Closed Circuit
Step 1
SpaceX / xAI need capital
$20.7B capex in 2025
xAI burning $2.5B/quarter
→
Step 2
High-interest bonds issued
Private debt at 8–12%
to fund xAI & Colossus
→
Step 3
Banks provide bridge loans
Morgan Stanley · Goldman
JPMorgan · BofA · Citi
at lower rates, secured vs IPO
↓
Step 6 — You are here
Public investors buy
$75B raised from retail
& institutions at IPO
←
Step 5
Banks underwrite IPO
Same banks now price
and sell the shares
Fee: ~$2.6B at 3.5%
←
Step 4
High-interest bonds retired
Bridge loans replace costly
private debt. Banks now
control the exit timing.
↓
Step 7 — Banks exit clean
IPO proceeds repay bridge loans + ~$2.6B in underwriting fees
Banks have earned: (a) bridge loan interest, (b) underwriting fees. Their credit risk is transferred to public investors.
The conflict of interest in plain English: The five banks are simultaneously the lenders (who want the IPO to raise enough to repay their bridge loans) and the price-setters (who control what price the IPO is offered at). Their incentive is to price the IPO as high as possible — more proceeds repay more bridge debt and generate larger fees. There is no independent party representing what the stock is actually worth to a long-term holder. Public investors absorb the credit risk the banks assumed during the bridge phase. This is the "overdraft" of the valuation — the price is set to satisfy the financing chain, not to reflect fundamental value.
Bridge loan exposure (est.)
$15–20B
Banks' skin in the game
Underwriting fees (~3.5%)
~$2.6B
Collected regardless of performance
Revenue multiple at IPO
94–107×
Apple: 7×. Amazon: 3.8×.
valuation — the gap between story and numbers
Revenue multiple comparison — SpaceX vs. benchmarks
SpaceX at 94× revenue. Even Starlink at its fastest growth doesn't mathematically close this gap for 10+ years.
day 1 — five scenarios and how to react
Scenario
Move
What's driving it
Your move
MOON
+40–80%
Retail FOMO + short squeeze + Musk hype. No new information.
Do not chase. History shows +80% Day 1 IPOs (Airbnb, DoorDash) fell 50–75% within 18 months. Wait for the retest.
STRONG
+15–30%
Healthy demand. Institutional buyers visible. Starlink narrative lands well.
Consider a small starter position (25% of intended). Hold rest for lock-up expiry dip in ~180 days.
FLAT
0–10%
Institutional flipping. xAI losses dominate headlines. S&P 500 exclusion noted widely.
Best near-term entry point. Price reflects reality, not euphoria. Build full position here.
WEAK
-5–15%
Soft book-build. OSHA investigation headline. Macro event. xAI scrutiny.
Banks will stabilise near IPO price for 30 days (green shoe). Use that window to accumulate in tranches.
BREAK
-20%+
Macro crash week of IPO. Legal action on Grok. OSHA shutdown order. Musk controversy.
Wait for one full session of stabilisation before entering. Do not buy a freefall. Diagnose cause first — operational vs. sentiment.
Note: S&P 500 requires 4 consecutive quarters of GAAP profit — SpaceX is ineligible at listing. No passive index fund buying on Day 1.
the verdict
Should you participate?
What is genuinely good: Starlink is a near-monopoly satellite internet business growing 50% annually with $7.2B EBITDA. The Anthropic compute contract proves real enterprise demand. The Colossus data centre, if the power and talent problems are solved, addresses the largest growing cost line in enterprise technology — AI compute.
What is genuinely concerning: At $1.75 trillion, you are paying 94× revenue for a company reporting a $4.94 billion annual loss. The underwriting banks have a structural incentive to overprice the IPO. The Anthropic contract is not locked in and can exit in 90 days. The orbital AI vision is 2+ years from generating a dollar of revenue. Nine of eleven xAI co-founders have departed. The power grid, GPU supply chain, and funding cycle all face hard constraints that cannot be solved by ambition alone.
The one-sentence verdict: Starlink alone is worth owning — but at $1.75 trillion, you are paying for three other businesses you may not want, at a price set by banks who need the IPO to succeed more than they need you to profit. If you participate, size small, expect volatility, and treat the lock-up expiry (180 days post-IPO) as your real entry point.
For informational purposes only. Not financial advice. Data from SpaceX S-1 (SEC EDGAR CIK 0001181412), filed May 20, 2026. All financial figures from public filings. Capital loop analysis is an interpretive framework, not a legal characterisation of the underwriting process. Consult a qualified financial advisor before making investment decisions.
PERSONAL RESEARCH · 2026 · SPCX