The Trillion-Dollar Month: How AI's Biggest Players Cashed Out, Built Empires, and Outran the Law

June 2026 might end up being the month historians point to when they explain how the AI industry stopped being a Silicon Valley story and became a Wall Street one. In the space of three weeks, two companies most people interact with every day  and one most people have never directly used lined up to become public companies at valuations that would have sounded like satire two years ago. At the same time, the laws meant to govern how AI gets used quietly lost a race against their own deadlines.

Here is what actually happened this month, who is paying for it, and what the receipts say.

The IPO Queue Nobody Saw Coming

Three companies. Three filings. A combined target valuation north of $4 trillion. That is the headline of June 2026.

SpaceX went first, and it went big. On June 11, Elon Musk's company  which absorbed his AI venture xAI earlier this year priced its initial public offering at a fixed $135 a share, raising $75 billion and pushing its valuation to roughly $1.75 trillion. NPR called it the biggest initial public offering on record, and the first of a trio of mega-IPOs from AI companies expected in 2026. The company said in its SEC prospectus that it would use the proceeds to grow its AI compute infrastructure, launch facilities, and satellite constellations. Source: NPR

Not everyone is convinced the price tag matches reality. Morningstar published a note calling SpaceX significantly overvalued and pegging fair value closer to $780 billion less than half the IPO target, largely because Starlink remains the only consistently profitable part of the business while the xAI unit's path to profitability is unclear. Source: CNBC An independent valuation breakdown by FutureSearch reached a similar conclusion from a different angle, estimating SpaceX's sum-of-the-parts fair value at around $1.25 trillion, about 29 percent below the IPO price. Source: FutureSearch

Anthropic moved next, and arguably made the bigger statement. On June 1, the Claude developer confidentially filed a draft S-1 with the SEC, days after closing a $65 billion funding round that valued the company at $965 billion. Bloomberg confirmed the round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with each lead investor putting in more than $2 billion, and reported that the deal eclipsed rival OpenAI's valuation for the first time. Source: Bloomberg TechCrunch added that the round also pulled in infrastructure-side strategic partners including Samsung, SK Hynix, and Micron chipmakers, not just financiers, betting directly on Anthropic. Source: TechCrun

The growth numbers behind that valuation are genuinely unusual. Fortune reported Anthropic's revenue run-rate hit roughly $47 billion, up from about $9 billion the year before an eighty-fold increase in annualized revenue. Source: Fortune But a more skeptical read from independent analysis site Humai points out that a meaningful chunk of that headline revenue figure consists of money Anthropic pays straight back to its own cloud partners, Amazon and Google a detail that matters when investors try to figure out what the company's real margins look like. Source: Humai Not every investor is sold on the trillion-dollar framing either Reuters and Business Insider reported that investor Michael Burry warned publicly that there's "no guarantee" Anthropic gets close to a $1 trillion price tag, calling frontier AI model-building "far too expensive." Source: TS2/Business Insider

OpenAI is reportedly right behind them. Multiple outlets reported the company is preparing its own confidential S-1, with Bloomberg sources indicating it is working with Goldman Sachs and targeting a public debut later this year. Source: Yahoo Finance/Bloomberg Intelligence

Why does the order matter? Because, as Reuters and IPO-tracking analysts have noted, whichever company lists first sets the market's benchmark for how investors price the entire category. If SpaceX's debut holds up, Anthropic and OpenAI ride the tailwind. If it doesn't, both face a tougher room.

Who Is Actually Footing the Bill

Here's the part that doesn't make the headline but probably should: none of this is being paid for out of pocket. It's being financed, leveraged, and packaged the way Wall Street finances anything large through debt, strategic partnerships, and government-adjacent capital.

The clearest example is also one of the strangest financial structures to emerge from the AI boom. To fund the chips and compute Anthropic needs, Apollo Global Management and Blackstone arranged a $36 billion debt deal but structured it so Anthropic never has to put the debt on its own balance sheet. A special-purpose vehicle holds the debt, buys the hardware (Google TPUs, in this case), and leases the capacity back to Anthropic. The financing site BuildFastWithAI summarized the strategic logic plainly: Anthropic gets the compute it needs without the debt liability, while Apollo and Blackstone collect yield on debt backstopped by Broadcom's balance sheet. Source: Build Fast with AI

Then there's the geography of it. On May 30, SoftBank announced plans to invest up to €75 billion (roughly $87 billion) to build 5 gigawatts of AI data center capacity in France the company's official release calling it its largest AI infrastructure investment in Europe, with the first €45 billion phase delivering 3.1 GW of capacity in the Hauts-de-France region by 2031. Source: SoftBank Group official release French Economy Minister Roland Lescure framed it as proof of President Macron's ambition to position France as a leading destination along the entire AI value chain, and state utility EDF is supplying a former nuclear power plant site for one of the campuses. Source: TechCrunch CNBC reported the deal arrived specifically because Europe's high energy costs have become a major obstacle in its bid to compete with the US and China on AI, and data center investment is now following whichever country has the cheapest, most reliable grid. Source: CNBC

That detail power availability now functioning as the deciding factor in where AI infrastructure gets built was echoed almost word for word by a former US Department of Energy official quoted in Data Center Knowledge's coverage of the same deal, who called power availability the primary site-selection variable for large-scale AI infrastructure investment today. Source: Data Center Knowledge So when a blog headline asks who is paying for the AI boom, the honest answer is layered: private credit funds collecting yield, sovereign governments offering cheap power and tax incentives, chipmakers buying equity instead of just selling hardware, and retail investors who will inherit the volatility once these IPOs actually list.

The Law That Blinked First

While the financial markets were busy setting records, something quieter and arguably more consequential for ordinary users happened in Colorado.

The Colorado AI Act was supposed to be the first comprehensive US state law governing how AI is used in decisions about employment, housing, healthcare, and credit. It had already survived one delay, pushed from February to June 30, 2026. Multiple law firms had spent the spring telling clients to prepare for that June 30 deadline.

Then, on May 14, 2026, Colorado Governor Jared Polis signed SB 189, which according to law firm Hunton's compliance bulletin revised the law and delayed its effective date again, this time from June 30, 2026, to January 1, 2027, while significantly scaling back its original requirements. The same bulletin noted the rewrite eliminates the duty of care aimed at preventing algorithmic discrimination, deployer risk-management obligations, and certain reporting requirements to the Colorado Attorney General, replacing the original risk-based framework with a narrower disclosure-and-transparency regime. Source: Hunton Andrews Kurth

This matters beyond Colorado. Law firm Baker McKenzie's tracking of the case noted that even before the rewrite, a federal court had already paused enforcement of the original law pending the state's rulemaking process, after a major tech company sued arguing the law was unconstitutional. Source: Connect On Tech / Baker McKenzie Meanwhile, at the federal level, Congress has its own 269-page "Great American AI Act" sitting in committee, which would preempt state AI laws like Colorado's for three years if it ever passes but as of mid-June, it had not moved out of committee, leaving companies who bet on federal preemption instead of state compliance in an awkward spot. Source: Build Fast with AI, June 5 2026 coverage

Put together: the most ambitious AI accountability law in the US just got weaker and later, right as the industry it was meant to regulate is about to start trading on the Nasdaq.

So What Does This Month Actually Tell Us

Three things, and none of them are subtle.

First, the AI industry has fully entered its financialization phase. When a chip-debt deal needs an SPV to keep liabilities off a balance sheet, and a national government negotiates a data center deal personally between a head of state and a company founder, this is no longer venture-funded experimentation it's sovereign-scale infrastructure finance.

Second, the people putting up the actual cash are increasingly not the companies themselves. It's private credit (Apollo, Blackstone), sovereign-adjacent capital (SoftBank, French state utility EDF), and public markets about to absorb three trillion-dollar-scale floats in the same calendar year which analysts at Goldman Sachs and elsewhere have flagged as a real capacity constraint, not just a metaphor.

Third, regulation is structurally incapable of moving at the speed the industry is moving. Colorado's law took two years to write and got delayed twice before it ever took effect. By the time any AI-specific law actually bites, the companies it targets may already be publicly traded, globally distributed across three continents, and financially entangled with the very governments meant to police them.

Whether SpaceX's stock holds its $1.75 trillion price, whether Anthropic's S-1 reveals margins healthy enough to justify $965 billion, and whether Colorado's law survives its next rewrite

those are the threads worth watching for the rest of 2026. This month just set the stakes.

FAQs

1. Why did Anthropic, OpenAI, and SpaceX all file for IPOs around the same time?

None of them coordinated it directly, but analysts point to the same underlying pressure: AI compute costs have grown so large that private funding rounds, even ones in the tens of billions, are no longer enough to cover the bill. Going public opens access to a much bigger pool of capital. Anthropic also had a strategic incentive  filing before OpenAI let it set the benchmark for how the market values a frontier AI company.

2. What is SpaceX's actual IPO valuation, and is it considered fairly priced?

SpaceX priced its IPO at $135 a share on June 11, 2026, implying a valuation of roughly $1.75 trillion. That price is disputed: Morningstar's independent valuation put fair value closer to $780 billion, and FutureSearch's segment-by-segment analysis estimated around $1.25 trillion both well below the IPO price.

3. How much is Anthropic actually worth, and is the $965 billion figure reliable?

Anthropic's $965 billion figure comes from its May 2026 Series H funding round. It's a private-market valuation, not yet a public trading price, so it reflects what large institutional investors agreed to pay not necessarily what public markets will pay once shares trade freely.

4. Is Anthropic's $47 billion revenue figure misleading?

It's accurate but incomplete on its own. A significant portion of that revenue is paid straight back out to cloud infrastructure partners like Amazon and Google, which affects how investors should read the company's real profit margins ahead of an IPO.

5. Who is actually funding the AI infrastructure boom the AI companies themselves, or someone else?

Increasingly, someone else. Private credit firms like Apollo and Blackstone are structuring multi-billion-dollar debt deals to buy chips on AI companies' behalf, often through special-purpose vehicles that keep the debt off the AI company's own balance sheet. Governments and state utilities, like France's EDF, are also contributing land, power infrastructure, and incentives.

6. Why is SoftBank investing €75 billion in French data centers specifically?

France offered three things AI infrastructure needs most: access to a reliable, low-carbon electrical grid, available industrial land, and a skilled engineering workforce. Europe's high energy costs have made site selection a major bottleneck for AI data centers generally, and France's grid stood out as a competitive advantage.

7. Is the Colorado AI Act in effect right now?

No. Despite earlier expectations that it would take effect on June 30, 2026, Colorado Governor Jared Polis signed SB 189 on May 14, 2026, delaying the law's effective date to January 1, 2027, and significantly narrowing its requirements.

8. What did the original Colorado AI Act require, and what changed?

The original 2024 law required AI developers and deployers to use "reasonable care" to prevent algorithmic discrimination in high-risk systems, including impact assessments and risk-management programs. The May 2026 rewrite removed most of those obligations and replaced them with a narrower framework focused mainly on disclosure and transparency for automated decision-making tools.

9. Could a federal AI law override state laws like Colorado's?

That's the goal of the proposed "Great American Artificial Intelligence Act," a 269-page federal bill that would preempt state AI laws for three years if passed. As of mid-June 2026, however, the bill remained stuck in committee and had not become law, so it has not yet superseded any state regulation.

10. What should everyday investors take away from this month's AI IPO wave?

Mainly caution paired with attention. Multiple independent valuations of SpaceX, and skepticism from high-profile investors like Michael Burry regarding Anthropic, suggest a real gap between private-market pricing and what public markets may ultimately support. None of this is investment advice — it's a reminder to read the actual S-1 filings once they become public, rather than relying on valuation headlines alone.

---

Reporting and analysis by Muntazir Mahdi, Founder of ANFA Technology and CS scholar at the University of Karachi. International verification and fact-checking by Mohsin Raza, International Relations researcher at Corvinus University of Budapest.

All figures, quotes, and claims in this article are sourced and linked directly to original reporting from Bloomberg, TechCrunch, Fortune, NPR, CNBC, Reuters-sourced coverage, and official company/government statements as cited above. This article reflects publicly available information as of June 21, 2026, and may be subject to change as IPO pricing, legal proceedings, and regulatory text evolve.