The AI IPO Race Anthropic, OpenAI & xAI - Who Goes Public First?

Yanik•03/22/2026•13 min read

Three companies. Up to $3.1 trillion in combined market cap. All trying to go public in the same year.

Anthropic, OpenAI, and Musk's freshly merged SpaceX-xAI are heading for what could be the largest synchronized liquidity event the public markets have ever absorbed. If you add up every VC-backed IPO since 2000, these three might still create more value.

None of these are victory-lap IPOs. They are a sprint for public capital to feed the compute furnace. The infrastructure costs of reaching AGI have outpaced what private markets can provide. OpenAI is projecting $17 billion in cash burn this year. Anthropic is in the middle of a $50 billion data center push. Musk's xAI needs SpaceX's balance sheet to stay solvent. Going public is not a strategic choice anymore. It is the only way to fund the energy and hardware bills at the current frontier.

What makes this particularly strange is that all three are betting on incompatible visions of what AI should be. Enterprise safety and disciplined scaling. Consumer ubiquity and platform lock-in. One man's attempt to merge rockets, social media, and artificial intelligence into a trillion-dollar conglomerate. The public market will judge all three, probably within months of each other.

The scoreboard

	Anthropic	OpenAI	SpaceX-xAI
Target valuation	$380B	~$1T (target, currently $730B pre-money)	$1.25–1.5T (combined)
Total funding raised	~$64B	~$150B	~$52B+
Annualized revenue	$19B+ (confirmed Mar 2026)	$25B (run rate)	~$500M (xAI) + $15–16B (SpaceX)
2026 cash burn	Not disclosed	$17B projected	xAI heavily cash-negative; SpaceX $8B profit
Expected filing	2026 (Wilson Sonsini engaged)	H2 2026 / 2027	July 2026 (S-1 reportedly filed)
Corporate structure	Public Benefit Corporation	For-profit PBC (converted)	Merged entity
Competitive moat	Enterprise safety, 1M token context	AI OS, Stargate infrastructure	Vertical integration, orbital compute
Key backers	Google, Amazon, Coatue, GIC	Microsoft, SoftBank, Thrive	Nvidia, Fidelity, QIA, Tesla

Anthropic - The safety premium

In February 2026, Anthropic closed a $30 billion funding round at a $380 billion post-money valuation, more than double its September 2025 number and the second-largest private financing round in tech history. The round was led by Coatue and Singapore's GIC, with Microsoft, Nvidia, D.E. Shaw, Dragoneer, Founders Fund, Iconiq, and MGX also participating. The five-year-old company has now raised nearly $64 billion total.

The valuation is not pure speculation. Annualized revenue hit $14 billion at the time of the February round and has since climbed to $19 billion run-rate, confirmed by CEO Dario Amodei at a Morgan Stanley TMT conference in early March. The company exited 2025 at a $9 billion pace and is expected to more than double to $30 billion by end of 2026. Epoch AI's analysis suggests Anthropic could surpass OpenAI in annualized revenue by mid-2026. Enterprise subscriptions quadrupled in early 2026.

Where the revenue comes from

Anthropic built its business almost entirely on enterprise and developer adoption, deliberately avoiding the consumer-first model. The flagship product: Claude Code, an AI coding agent that alone generates $2.5 billion in annualized revenue. Eight of the Fortune 10 are Claude customers. Over 500 companies spend more than $1 million a year on the platform.

The technical moat is concrete. Claude Opus 4.6 has a 1-million-token context window, roughly 75,000 lines of code in a single pass. This is not a chatbot feature. It is an enterprise infrastructure capability, and it has contributed to a $2 trillion market cap loss for legacy software stocks as companies replace traditional tooling with AI-native alternatives.

Anthropic's pitch to public investors writes itself: recurring revenue, deep integrations, disciplined spending, and customer stickiness that translates to predictable cash flows. Fewer viral consumer moments than ChatGPT, but significantly higher revenue per customer.

Infrastructure push

Behind the revenue story is a massive buildout. Anthropic is spending $50 billion on data centers, anchored by an $11 billion AWS campus and a strategic automation partnership with Fluidstack. This is the capital that needs funding, and the primary driver behind the IPO calculus.

The IPO question

Anthropic has engaged Wilson Sonsini for IPO-related preparations, placing it firmly in the 2026 listing discussion. But there is no S-1, no confirmed underwriter, and no roadshow date. Insiders describe these as early-stage conversations.

At $380 billion private valuation with double-digit-billion revenue, Anthropic is already valued like a large-cap public company. The longer it stays private, the more it depends on secondary markets and tender offers to provide liquidity to employees and early investors.

Risk: Pentagon blacklist

Anthropic was designated a "supply chain risk" by the Pentagon after refusing to let its models be used for mass surveillance or autonomous weapons. This aligns with the company's safety brand, but it creates a ceiling for government procurement revenue that rivals like OpenAI, who took the Pentagon money, don't face. ESG-conscious funds will reward the stance. Defense-heavy portfolios will penalize it.

Risk: Valuation multiple

At $380B on $19B revenue, Anthropic trades at 20x revenue in the private market. That is aggressive even for hypergrowth SaaS. If the broader AI trade corrects before Anthropic files, or if revenue growth decelerates, the IPO window could narrow fast.

OpenAI - Scale at all costs

OpenAI is the name everyone knows. ChatGPT has 910 million weekly active users. Over 9 million businesses pay for it. The company hit $25 billion in annualized revenue by February 2026, a 233% year-over-year increase, making it the largest AI company by top line. With roughly $150 billion in total funding raised, it is also the most capitalized private company in history.

And yet OpenAI's path to an IPO is arguably the most complicated of the three.

The restructuring

The single biggest development in OpenAI's corporate history: the company completed its conversion from a nonprofit with a for-profit subsidiary into OpenAI Group PBC, a traditional for-profit public benefit corporation. The nonprofit OpenAI Foundation remains in control with a ~$130 billion equity stake. This was a prerequisite for any IPO, since public investors cannot buy shares in a nonprofit. The Wall Street Journal reported that some recent funding was conditional on the conversion being completed by end of 2025.

The structural barrier to going public is gone. But so is the governance structure that once made OpenAI unique: the nonprofit board that could, in theory, prioritize humanity's interests over shareholders'.

The financial reality

Here is where things get uncomfortable. OpenAI projects $17 billion in cash burn for 2026. Despite $25 billion in revenue, the company's finances are burdened by structural obligations most investors haven't fully priced in:

A renegotiated Microsoft revenue share (October 2025) requiring 20% of total revenue through 2032, which is $5 billion per year at current rates, growing as revenue scales
A $250 billion lifetime commitment to Azure cloud services
Gross margins around 40%, constrained by compute costs that scale with every ChatGPT query

OpenAI's own internal forecasts project cumulative losses of $115 billion through 2029 before reaching profitability "sometime in the 2030s." The bull case: $100 billion in annual revenue by 2029. The bear case: a company burning $17 billion a year with a 20% Microsoft tax that needs constant capital infusions just to survive. That is a brutal story for public market investors who watched the SPAC era collapse.

The Stargate bet

OpenAI is also building physical infrastructure at a scale that rivals national energy projects. The $500 billion Stargate initiative is a 10-gigawatt data center consortium with SoftBank, Oracle, NVIDIA, and Crusoe. It is the most expensive private infrastructure project in history, and it is OpenAI's bet that whoever controls the most compute wins the AGI race.

IPO timeline

The company hired former DocuSign CFO Cynthia Gaylor as head of investor relations, a clear IPO signal. Internal targets point to filing in H2 2026 with a 2027 listing, though CFO Sarah Friar has suggested 2027 is more realistic. The latest $110 billion round valued OpenAI at $730 billion pre-money, implying an IPO target approaching $1 trillion.

Bull case

910M weekly users. $25B ARR growing at 233% YoY. The Stargate infrastructure moat. The Gaylor IR hire signals professional governance ahead of listing. If OpenAI can show a credible path to $100B revenue, the IPO could price on a forward multiple similar to early Amazon.

Risk: The loss problem

$17B projected cash burn in 2026. 20% Microsoft revenue share through 2032. $250B Azure commitment. $115B cumulative losses through 2029. No profitability until the 2030s. After WeWork and the SPAC implosion, "growth at all costs" is a harder sell than it was in 2021.

xAI / SpaceX - The Musk gambit

On February 2, 2026, Musk did something nobody expected: he merged the world's leading orbital launch provider with a two-year-old AI startup. SpaceX became the "managing member" of X.AI Holdings, combining rockets, satellite internet, and AI under one roof. Combined valuation: $1.25–1.5 trillion. SpaceX at $1 trillion, xAI at $250 billion. Strategic investors include NVIDIA, Cisco, Valor Equity Partners, Fidelity, Qatar Investment Authority, MGX, and a $2 billion injection from Musk's own Tesla. The entity has reportedly filed its S-1, targeting a public debut as early as July 2026 at $1.5 to $1.75 trillion.

On paper, SpaceX-xAI is first to file. In practice, the story is deeply fractured.

The exodus

Nine of xAI's eleven original co-founders have left. As of March 2026, only two remain. The departures include Tony Wu, Zihang Dai, Guodong Zhang, Jimmy Ba, and Toby Pohlen, researchers central to Grok's early architecture and training. This is not normal startup attrition. This is a gutting of the founding technical team.

Musk acknowledged it publicly on March 13: xAI "was not built right first time around, so is being rebuilt from the foundations up." That admission, weeks after Tesla invested $2 billion into xAI, is candid and damaging in equal measure. When the founder of your AI division says publicly that it needs to be rebuilt from scratch, the $250 billion valuation attached to it gets very hard to defend.

Product problems

Grok, xAI's flagship chatbot, faces headwinds on every front. Musk has publicly admitted that "Grok is currently behind competitors in coding," which is the single most important enterprise AI benchmark and exactly where Anthropic's Claude Code and OpenAI's tools dominate. On the regulatory side, Grok enabled users to generate non-consensual sexual imagery, including deepfakes of real people and minors, triggering government investigations in multiple jurisdictions. And the product is primarily distributed through X (formerly Twitter), limiting reach compared to ChatGPT's standalone distribution or Claude's enterprise API.

The SpaceX subsidy

The merger's logic is financial, not technical. xAI's ~$500 million in 2025 revenue (projected to reach >$2 billion in 2026) cannot sustain a $250 billion valuation independently. SpaceX provides the foundation: roughly $8 billion in profit on $15–16 billion in revenue, a 50% net margin. That cash flow is what subsidizes xAI's development and the "Colossus" compute buildout.

For IPO investors, the question is simple: how much of the $1.5–1.75 trillion target is SpaceX (proven, profitable) and how much is xAI (a company whose founder says needs rebuilding)? If the market assigns even modest scrutiny to the xAI portion, the premium over SpaceX's standalone value disappears.

The orbital compute thesis

The long-term bull case for xAI centers on space-based AI compute. Using Starlink's 9,000+ satellite constellation and SpaceX's launch economics, Musk has said the goal is achieving the world's lowest-cost AI compute by putting data centers in orbit within two to three years. The ultimate vertical integration play: controlling the launch vehicle, the satellite network, and the AI workload.

Is this visionary infrastructure planning or sci-fi marketing? Hard to say. What is clear: no other AI company can even attempt it.

Risk: Talent and product

9 of 11 co-founders gone. Founder admits it "was not built right." Grok behind in coding benchmarks. Government investigations in multiple jurisdictions. xAI revenue at ~$500M against a $250B valuation.

Bull case

SpaceX is SpaceX. Orbital launch dominance, Starlink approaching profitability at scale, 50% net margins, stable government contracts. If investors underwrite SpaceX at $1T+ and treat xAI as a free option on orbital compute upside, the IPO works, but on the strength of rockets, not chatbots.

The float paradox: when $3 trillion hits the market

Step back and think about what is about to happen to the plumbing of the financial system.

At standard float percentages of 15–20%, these three firms would need the market to absorb between $432 billion and $576 billion in new equity. The entire U.S. IPO market from 2016 to 2025 raised $469 billion. Three companies, in a single year, would require as much capital as the entire American IPO market generated over the last decade.

This is what analysts are calling the Float Paradox. To maintain founder control and avoid crashing the market, these firms will almost certainly debut with tiny 3–8% public floats. But S&P 500 inclusion requires a 50% public float. That gap has three consequences worth tracking:

1. Index exclusion at listing

Despite trillion-dollar valuations, all three will initially be excluded from the S&P 500. The largest pool of automatic buyers, passive index funds managing $20 trillion, will not be purchasing at IPO. Demand at listing comes entirely from active managers and retail.

2. The delayed forced liquidation

When these companies eventually reach the 50% float threshold (through secondaries, lockup expirations, or insider sales), passive funds will be required to buy. To raise the cash, they will sell existing positions, primarily in current mega-caps like Apple, Microsoft, and Nvidia. The scale of forced selling needed to absorb three new trillion-dollar entrants into the index has no precedent.

3. The momentum cascade

Massive selling in current market leaders to fund new entrants triggers algorithmic momentum sell-offs, creating downward pressure across the broader index at the exact moment these new AI stocks are being added. The act of including the 2026 AI IPOs in the S&P 500 could destabilize the companies that currently define it.

The math

Up to $3.1 trillion in combined market cap. At 15% eventual float: $435 billion in shares requiring placement. Passive funds managing $20 trillion must rebalance to absorb them. Even a 2% portfolio reallocation across passive funds means $400 billion in forced sales of existing holdings. This is a stress test for the entire architecture of passive investing.

The defense contract culture war

There is a strategic divergence between these three companies that most coverage ignores: their relationship with the U.S. military.

OpenAI took the Pentagon money. After years of internal debate and a 2018 pledge never to build weapons, the company reversed course and signed defense contracts, including an arrangement highlighted by Sacra in early 2026. Pragmatic play: defense revenue is sticky, high-margin, and buys political cover when geopolitical tensions are rising.

Anthropic drew red lines. The company refused to let its models be used for mass surveillance or autonomous weapons and paid the price. The Pentagon designated Anthropic a "supply chain risk," effectively blacklisting it from major defense procurement. Brand-consistent for a company built on "responsible AI." For its IPO, it creates a ceiling on government revenue that rivals don't face.

xAI/SpaceX is the most deeply embedded. SpaceX already holds critical national security launch contracts. The merger positions the combined entity as a potential end-to-end defense AI provider: satellites, data, and intelligence models under one roof. The most aggressive government play of the three, but one that ties the company's fortunes to the political winds of defense spending.

For investors, this is not abstract policy. It is a revenue segmentation question. Government AI spending is projected to exceed $100 billion annually by 2028. How each company positions itself against that revenue pool will materially affect their public market performance.

The BearBull take: a $3 trillion game of chicken

The 2026 AI IPO race is not a victory lap. It is a sprint for public liquidity to fund a cash furnace that private markets can no longer feed.

OpenAI is a $730 billion company projecting $17 billion in cash burn this year. Add a 20% revenue share to Microsoft through 2032 and a $250 billion Azure commitment: they are not going public to reward early investors. They are going public to survive the cost of their own infrastructure. Stargate is a $500 billion bet that whoever controls the most compute wins, but it only works if public markets keep writing checks.

Musk has slapped a $1.25–1.5 trillion price tag on a Frankenstein merger between SpaceX and xAI. The orbital compute idea is interesting, nobody else can put data centers in space, but the reality on the ground is ugly: xAI's founders are fleeing, Grok is drowning in deepfake regulatory probes, and Musk has openly said the company needs to be rebuilt from the studs. SpaceX is world-class. xAI is a $250 billion question mark.

If there is a winner on fundamentals, it is Anthropic. By positioning as the adults in the room, rejecting defense surveillance contracts, focusing on enterprise reliability, scaling to $19 billion ARR without the existential drama, they have built the cleanest financial story of the three. The quadrupling of enterprise subscriptions and Claude Code's $2.5 billion standalone ARR are hard numbers, not projections. Anthropic's risk is valuation, not viability.

But the real story for investors is market mechanics. These three will drop nearly $3 trillion in market cap onto the public markets. They will IPO with microscopic floats to avoid draining liquidity, and their eventual S&P 500 inclusion will trigger a violent, mandatory sell-off of legacy tech giants by index funds. The 2026 AI IPO wave will make a few insiders astronomically rich. It might also break the plumbing of the stock market.

Disclaimer

This article is for informational and educational purposes only. Nothing in this article constitutes investment advice, financial advice, or a recommendation to buy or sell any security. All data, figures, and projections are sourced from publicly available information and may be incomplete or outdated. Investing involves risk, including the possible loss of principal. Always conduct your own research and consult a licensed financial advisor before making investment decisions.