SpaceX SPCX : +19% on first day of trading, Musk becomes the first trillionaire — AI accelerates the space bubble
🔎 June 12, 2026: the IPO that shattered all records
On June 12, 2026, SpaceX opened trading on the Nasdaq under the ticker SPCX at $150, to close at $161 — a 19% increase over the IPO price set at $135 the day before.
This is the largest initial public offering in history: 555 million shares issued, $75 billion raised, and a valuation exceeding $2,100 billion at the end of the session according to The Guardian.
Elon Musk, with his combined stakes in SpaceX and Tesla, officially becomes the first trillionaire in human history. Forbes estimates his fortune at $1,100 billion at the close.
But behind the media spectacle, there is a question that very few commentators are really asking: what exactly is SpaceX selling to justify $2,100 billion?
The key points
- SPCX opens at $150, closes at $161$, +19% on the IPO price of $135$ — largest IPO of all time ($75B raised).
- Musk becomes the first trillionaire in history, fortune estimated at $1,100B by Forbes.
- Closing valuation: ~$2,100B, 6th largest US market capitalization according to Reuters.
- Only profitable division: Starlink. The rest (launchers, Starship, AI compute) burns cash.
- The real driver of the valuation is no longer space-related — it's GPU infrastructure (COLOSSUS, agreements with Google and Anthropic).
- Morningstar estimates the fair value at about half the closing price.
Recommended tools
| Tool | Main usage | Price (June 2026, check on site.com) | Ideal for |
|---|---|---|---|
| Hostinger | Web hosting to follow tech news | 2.99$/month | Creating a financial/tech blog |
| Meilleurs outils IA | Automated market monitoring | Variable | Analysts and investors |
| Tendances IA | Tracking AI bubble/innovation cycles | Free | Understanding the AI-space link |
Today's figures: record volatility and market mechanics
From $135 to $161: the details of a historic day
SPCX stock experienced a roller coaster day. The intraday range oscillated between $149 and $176, according to the live coverage of the New York Times.
This type of volatility is typical of hype IPOs, but the amplitude is unusual for a company of this size. For comparison, Alibaba's IPO in 2014 had shown a much tighter range.
The close at $161 represents a 19% premium over the IPO price. According to Business Insider, this performance surpasses that of Meta (2012), Alibaba (2014), and even Tesla (2010) on their first day.
MSCI inclusion: the hidden catalyst of June 13
As of June 13, SPCX enters the MSCI USA indices. This is a crucial point that many superficial comments ignore.
Index funds and ETFs that replicate the MSCI USA will mechanically have to buy between $8 and $12 billion worth of SPCX to maintain their weighting. This forced buying creates an artificial demand floor in the weeks following the IPO.
This phenomenon is not new — we saw it with Tesla when it joined the S&P 500 in December 2020. But the scale is unprecedented.
SpaceX's financial structure: what are the different divisions really worth?
Starlink: the only cash cow
According to the S-1 analyzed by The VC Corner, Starlink is the only SpaceX division that generates a net operating profit. The satellite segment has several million subscribers worldwide and is growing at a double-digit rate.
But even projecting optimistic growth over 10 years, Starlink alone does not justify $2.1 trillion. Most independent DCF models arrive at a valuation between $400 and $800 billion for this division alone.
Launchers and Starship: prestige, not profits
The launcher business (Falcon 9, Falcon Heavy) is mature but has low margins. Competition from Blue Origin and Rocket Lab is intensifying.
Starship, the heavy launcher, is still in the development phase. Each test launch costs hundreds of millions. Its business model (Martian colonization, launching mega-constellation satellites) relies on hypotheses that remain to be proven.
AI compute: the real speculative bet
This is where the narrative gets interesting. SpaceX is no longer positioning itself solely as a space company. It has become a massive AI infrastructure provider.
The COLOSSUS cluster, located on the same campus as SpaceX facilities, represents one of the world's largest supercomputers dedicated to training language models. To understand the scale of this transformation, read our article on SpaceX devient fournisseur d'infrastructure IA : Google paie 920 millions de dollars par mois pour 110 000 GPUs NVIDIA.
The AI link: why investors are buying SPCX for compute exposure
The xAI acquisition: SpaceX absorbs Grok
In February 2026, SpaceX acquired xAI for an estimated value of $125 billion, according to Wikipedia. This acquisition merged Grok's capabilities (including Grok 4.1, ranked 90/100 in general LLM) with SpaceX's physical infrastructure.
The S-1 reveals that Grok 5 is trained on the same campus as space operations. Sharing cooling, power, and fiber optics between AI and space significantly reduces marginal costs.
COLOSSUS: the supercomputer that justifies the premium
The VC Corner's analysis of the S-1 shows that SpaceX has converted compute into a massive enterprise contract. The COLOSSUS cluster serves to:
- Train Grok 5 (xAI, now internal).
- Subcontract for Anthropic (220,000 dedicated GPUs).
- GPU cloud contracts with Google, valued at $920 million per month.
This last figure is dizzying. At $920M/month, the Google contract alone generates ~$11B in annual recurring revenue. That is more than the entire launch business brings in.
The comparison with AI pure-plays
The paradox is striking. At a time when AI pure-plays like Anthropic or OpenAI remain private and burn through billions, SpaceX — a publicly traded company — offers indirect exposure to AI compute with the liquidity of a Nasdaq stock.
Institutional investors who cannot buy stakes in private Anthropic or xAI are falling back on SPCX. It is an AI proxy disguised as a space company.
Models trained on SpaceX infrastructure
What's running on COLOSSUS in June 2026?
The ecosystem of models that depends directly or indirectly on SpaceX infrastructure is considerable:
- Grok 4.1 (xAI/SpaceX): score of 90 in LLM general, 79 in agentic. In-house model, priority advantages on the cluster.
- Claude Opus 4.7 (Adaptive) (Anthropic): 94.3 in agentic, 90 in general. Anthropic outsources part of its training on COLOSSUS.
- GPT-5.5 (OpenAI): 98.2 in agentic, 91 in general. Although OpenAI has its own infrastructure, certain fine-tuning workloads go through third parties.
This infrastructure position gives SpaceX considerable market power. If Anthropic wants to scale Claude Opus 4.7 or its future versions, it depends partly on Musk's goodwill. This is a geopolitical lever as much as a commercial one.
For those who want to understand how these models are used in practice, our guide on creating your first autonomous AI agent details the agentic architectures that rely on these foundation models.
Musk trillionaire: the mechanics of wealth
How one goes from billionaire to trillionaire
Musk's wealth does not solely come from his direct holding of SPCX. According to CNBC, it is the combination of his stakes in SpaceX and Tesla that propels him beyond $1,000 billion.
The mechanism is simple in appearance: Musk holds about 42% of SpaceX shares (class B shares with multiple voting rights). At a $2,100B market cap, his stake is worth ~$880B. Add his Tesla stake (estimated at $200-250B in June 2026), and the total easily exceeds $1,100B.
The risks of concentration
The problem: this wealth is illiquid. Musk cannot sell his class B shares without losing control of SpaceX. And if he sold even a fraction of his Tesla shares, the stock price would collapse.
A trillionaire on paper, yes. But a trillionaire who cannot convert more than 5-10% of his wealth into cash without creating a market crisis.
Political reactions
The New York Times notes that the day was also marked by political reactions. The concentration of such wealth in the hands of a single person, who simultaneously advises the US government via DOGE, raises questions of conflict of interest.
These questions are not new, but the unprecedented scale of the figure ($1,000 billion) makes them impossible to ignore.
Historical analogies: bubble or new paradigm?
The Cisco 2000 parallel
In March 2000, Cisco Systems reached a market capitalization of $555 billion — the highest ever recorded at the time. The price/sales ratio exceeded 40x. This valuation was justified by the promise that "all internet traffic would go through Cisco".
Twenty-six years later, SpaceX is valued at $2.1T with a price/sales ratio likely above 30x (if we exclude GPU contracts which are still in the ramp-up phase). The promise: "all AI compute and all global internet access will go through SpaceX".
The narrative structure is identical. Only the sector changes.
The NVIDIA 2024-2025 parallel
NVIDIA showed that a company can justify an extreme valuation if its revenues grow at a sufficient pace. But NVIDIA had gross margins of 75%+ and a near-monopoly on training GPUs.
SpaceX, even with COLOSSUS, does not have this moat. Google, Microsoft, Amazon, and Oracle have their own GPU clusters. The $920M/month contract is significant, but it is neither exclusive nor guaranteed in perpetuity.
What is different this time
Two factors distinguish SpaceX from previous bubbles:
- Starlink generates real cash with positive operating margins. It is not a pre-revenue startup.
- Physical infrastructure is a real moat. Building a cluster of 220,000 GPUs with power and cooling takes 18-24 months. SpaceX has a head start.
But do these factors justify $2.1T? Morningstar does not think so.
Morningstar's take: fair value at half the price
Why fundamental analysts are skeptical
Morningstar, one of the most respected independent rating agencies, estimates that the fair value of SPCX is around $80-85 per share — roughly half of the first-day closing price.
Their reasoning rests on three pillars:
- Sum of the parts: even being generous with Starlink ($800B), AI compute ($400B), launchers ($200B), and Starship (option value: $100B), we reach a maximum of ~$1,500B.
- Future dilution: the S-1 plans for additional share issuances to fund Starship and the expansion of COLOSSUS.
- Regulatory risk: the de facto monopoly on satellite internet access and AI compute will attract antitrust attention.
Is the market right to ignore fundamentals?
Historically, markets can remain irrational longer than you can remain solvent (the famous Keynes quote). SPCX could very well rise to $200 or $250 in the coming weeks, driven by MSCI inclusion and FOMO.
But in the medium term (12-24 months), fundamental gravity always ends up taking over. To follow the evolution of these market dynamics, our AI trends page is updated monthly.
SpaceX's real business: rockets, Starlink, or AI compute?
Estimated valuation breakdown
Here is an approximate breakdown based on the S-1 data and independent analyses:
| Division | Estimated 2026 Revenues | Operating Profit | Implied Valuation | % of total |
|---|---|---|---|---|
| Starlink | ~$18B | ~$3.5B | ~$800B | 38% |
| AI Compute (COLOSSUS) | ~$15B | ~$1B | ~$700B | 33% |
| Launchers (Falcon) | ~$5B | ~$0.3B | ~$200B | 10% |
| Starship | ~$0.5B | Negative | ~$200B (option) | 10% |
| xAI/Grok | ~$0.5B | Negative | ~$200B (option) | 9% |
These figures are estimates based on the S-1 and analyses from The VC Corner. The "AI Compute" line includes contracts with Google and Anthropic.
The honest answer
SpaceX's real business in 2026 is a hybrid. Starlink brings the cash, AI compute brings the growth narrative, and rockets bring the brand.
But if you have to answer the question "why $2,100B", the answer is clear: investors are buying AI compute with a free space ticket. The Unitree GD01 piloted mecha at $650,000 illustrates another facet of this hardware-AI convergence, but it is SpaceX that reaps the greatest financial benefit from it.
The systemic effect: what the SPCX IPO changes for the AI market
A new listed proxy for AI
Until June 2026, investors who wanted pure AI exposure in public markets had two options: NVIDIA (GPU) or big tech (Microsoft, Google, Amazon via their clouds).
SPCX creates a third path: independent, listed compute infrastructure, with an additional narrative (space). It is an attractive financial product for asset managers who have to justify an "aggressive AI" allocation to their investment committees.
The pressure on private pure-plays
Anthropic, OpenAI, Mistral: all these companies will feel the pressure. If SpaceX can monetize compute at $920M/month via a contract with Google, why would investors accept funding operating losses in pure-plays?
The SPCX IPO could accelerate the IPOs of other AI infrastructure companies in 2027. The precedent is set: the market accepts aggressive valuations if the narrative is convincing enough.
The risk of increased correlation
The danger: if SPCX becomes an AI proxy, a correction in the AI sector will mechanically trigger a correction in SPCX — even if Starlink continues to grow. This cross-correlation did not exist when SpaceX was private.
❌ Common mistakes
Mistake 1: Confusing space valuation and AI valuation
Many commentators analyze SPCX as a pure space company. This is a mistake. If you remove AI compute from the equation, the fair value likely drops below $1,200B. Buying SPCX thinking you are buying "the future of space" means buying a disguised AI proxy.
The solution: analyze SPCX as a hybrid, and weight your investment thesis accordingly. If you don't want AI exposure, SPCX is not the stock for you.
Mistake 2: Ignoring future dilution
The S-1 anticipates additional share issuances. The diluted share count in 3-5 years will be significantly higher than today. A per-share valuation of $161 today does not mean $161 in three years, even if the total market cap remains stable.
The solution: look at the fully diluted market cap, not the price per share.
Mistake 3: Assuming the Google contract is guaranteed
$920M/month is huge. But cloud GPU contracts are regularly renegotiated. If Google decides to repatriate this workload to its own TPU clusters, SpaceX loses $11B in annual revenue overnight.
The solution: treat this contract as at-risk recurring revenue, not a guaranteed annuity.
❓ Frequently Asked Questions
What is the exact closing price of SPCX on June 12, 2026?
$161, or +19% compared to the IPO price of $135, according to CNBC and Business Insider. The intraday range was $149 to $176.
What is Elon Musk's exact net worth after the IPO?
Forbes estimates it at $1.1 trillion at the end of the day, combining his stakes in SpaceX (~$880B) and Tesla (~$220B). He becomes the first trillionaire in history.
Why is SpaceX worth more than $2 trillion?
The valuation relies on three pillars: Starlink (real cash flow), AI compute via COLOSSUS (strong growth, $920M/month Google contract), and Starship (optionality value). But Morningstar estimates the fair value at about half of that.
What is the connection between SpaceX and AI?
SpaceX hosts the COLOSSUS cluster, one of the largest in the world. xAI (acquired in February 2026) trains Grok there. Anthropic outsources part of the training of Claude Opus 4.7 there. Google rents GPUs there for $920M/month.
Will SPCX continue to rise after the first day?
The MSCI inclusion on June 13 creates mechanical buying of $8-12B by index funds, which could support the stock in the short term. But in the medium term, fundamental gravity (valuation vs. cash flows) could take over.
✅ Conclusion
SPCX at $161 is the story of a space company that convinced Wall Street it was primarily an AI company. The largest IPO in history rests on a paradox: the only profitable division (Starlink) does not justify half of the valuation, while the real speculative driver (AI compute) is competitive and reversible. Musk a trillionaire on paper, yes — but the real question is whether SPCX will be at $80 or $250 in 24 months. To follow the evolution of this type of dynamic, regularly check our ranking of the best AI tools.