Ultraman brought ChatGPT into the world, launching this AI revolution. But revolutions do not wait for anyone to get ready. Now, he is sitting on the back of a tiger that has already started running, with Anthropic, SpaceX, and Google in front of him, while behind him is a seemingly bottomless bill for computing power.

Written by: Wall Street View Research Team

In November 2022, ChatGPT went online, and the servers were overloaded that very day. Before this, no product from any tech company had ever gained 100 million users at such an astonishing speed.

After that moment, the "AI revolution" transformed from a term within the industry into a reality that ordinary people could feel.

OpenAI was the one that opened the curtain on this revolution.

Three years later, according to reports from The Information and other media, OpenAI is preparing to secretly submit its IPO draft to the SEC, aiming for a September listing with a target valuation of over 1 trillion dollars. Ultraman has not denied this report.

But just 48 hours ago, he told employees at a company meeting that submitting an IPO application and being truly ready to go public are two different things, and the company will not rashly enter the public market when the conditions are not ripe.

These two statements clarify OpenAI's current situation very clearly.

Rushing to the Market

Ultraman says he is not ready, but the company is reportedly set to file today. To understand this contradiction, one must first see the threefold pressure behind it.

First, Anthropic is pulling ahead.

In the past 15 months, Anthropic's annualized revenue has grown from 1 billion dollars to 30 billion dollars, a 30-fold increase. In the same period, OpenAI's revenue grew from about 20 billion dollars to 25 billion dollars, an increase of about 25%.

In April this year, Anthropic's annualized revenue officially surpassed that of OpenAI.

Worse figures appeared in Q2. Anthropic expects Q2 revenue to be about 10.9 billion dollars while achieving about 600 million dollars in operating profit, meaning Anthropic will turn a profit before OpenAI.

Anthropic's private valuation is already 1.2 trillion dollars, higher than OpenAI's own 1 trillion dollar IPO target.

Moreover, Anthropic is also preparing for an IPO, with the earliest date being October. Ultraman privately expressed a hope that OpenAI could go public first.

Second, SpaceX will be competing for funds next month.

Elon Musk's SpaceX plans to raise about 75 billion dollars at a valuation of 1.75 trillion dollars next month, which would be the largest IPO in history.

According to reports citing informed sources, there is "a lot of competition" between OpenAI and SpaceX, with OpenAI wanting investors to set aside money for it in advance, rather than putting all their resources into SpaceX.

The competition for public market funding between the two companies is real.

Musk was once a co-founder of OpenAI but left to create the rival xAI, and the two companies are currently involved in a lawsuit.

Now, Musk's other company is once again vying for resources from the same pool.

Third, the cost of computing power is a bottomless pit, and funds are running low.

OpenAI expects to spend 665 billion dollars on computing power by 2030. The company has raised nearly 200 billion dollars so far, but it is still far from that figure.

A public offering would be the next opportunity to supplement their resources, with no alternative options available.

With all three pressures combined, the logic becomes clear: it’s not that OpenAI is ready; it’s that it cannot wait.

Internal Disarray

This IPO lacks cohesive internal support.

Major investors are surprised by the pace of advancement, which is not a normal signal before an IPO.

OpenAI CFO Sarah Friar's attitude toward the IPO timeline is noticeably more cautious than Ultraman's, a point that has been reported, indicating real divisions within the company.

Ultraman himself mentioned at a company-wide meeting that the upcoming significant IPO by SpaceX and the global economic trends will be key external variables affecting whether OpenAI can go public. In other words, he himself is uncertain whether the September window can be opened.

This does not reflect the state of a company confidently marching toward an IPO. It resembles more a situation where the environment is pushing it forward, but even internally there is uncertainty about how far it can go.

Cracks in the Story

A deeper issue is that the story OpenAI tells to the public market is increasingly difficult to reconcile.

ChatGPT surged to fame, establishing OpenAI's unshakable brand advantage in the consumer space. However, in the past 12 months, ChatGPT's web traffic market share has declined from 87% to 68%, while Google Gemini's share has risen from 5% to over 18%, and the number of monthly active users has grown from 350 million to 750 million.

Google has search, Android, and Chrome—these entry points are things OpenAI does not have and cannot buy.

The enterprise side is also under pressure. In direct head-to-head competition, Claude has won about 70% of enterprise orders.

Anthropic has achieved a higher market revenue share with fewer users (134 million vs about 900 million for OpenAI) (31.4% vs 29%), and the average monthly revenue per active user is $16.2, leading in commercial efficiency.

The losses in these two directions are occurring within OpenAI's core base.

OpenAI's response is to bet on the AI Agent economy. If the AI Agent becomes the primary means of interaction between humans and machines, OpenAI's model is the operating system of this ecosystem, and the projected revenue of 280 billion dollars for 2030 is just a small part of the ecosystem.

This logic stands on a conceptual level. But it requires OpenAI to maintain growth across consumer, enterprise, and developer ecosystem on three fronts, while rivals in all three areas are accelerating, and OpenAI’s own growth rate is only 25%.

From 25 billion dollars to 280 billion dollars, in 4 years, at a 65% annual compound growth rate—this is the number the public market will want to verify.

The Costliest Reflection

This is the first time OpenAI must open its books.

Previously, the 852 billion dollar private valuation was built on a common narrative among a few institutional investors, with financial data being opaque and valuation methods inconsistent. Once the S-1 is made public, the amounts of losses, cost structures, reliance on Microsoft clauses, and the specific assumptions behind the 280 billion dollar forecast will all be laid bare in the public market.

The questions from the public market are quite direct: Can OpenAI win, when will it be profitable, and what justifies this price?

This is precisely the meaning of Ultraman's statement that "conditions are not yet ripe." He knows better than anyone that the private market buys stories, while the public market buys numbers. And OpenAI's current numbers—slowing revenue growth, declining market share, and continued losses—are not in the best shape for a public debut.

But it can no longer wait for the best state to appear.

Every passing quarter, Anthropic is narrowing the gap. With each day SpaceX advances, there is less money in the market. The bill for computing power is increasing every month.

This encapsulates the meaning of being "riding on a tiger": it is easy to climb on, but difficult to get down.

OpenAI has spent three years transforming itself into the biggest symbol of this AI revolution, and now this symbol has become a burden—it must be worthy of 1 trillion dollars, even if it is not yet ready.

Asset Impacts

Microsoft is the biggest beneficiary of this IPO. With an investment of 13.8 billion dollars corresponding to a 27% stake, if the IPO lands at a valuation of 1 trillion dollars, the unrealized gains would exceed 270 billion dollars. The short-term outlook is positive, and the buying logic is clear.

However, it should be noted that after OpenAI goes public, it will seek to reduce its dependence on Microsoft, which carries dilution risks for Microsoft's exclusive advantages in AI in the medium to long term.

NVIDIA: OpenAI is one of the biggest buyers of computing power, and raising funds from the IPO will accelerate computing expenditures, thus supporting GPU demand. However, in the medium to long term, OpenAI has a self-developed chip roadmap, which is a variable that needs to be discounted in the valuation.

CoreWeave: OpenAI's computing partner, which directly benefits from the acceleration of computing expenditures, can serve as a proxy target.

A-shares AI sector: If the 1 trillion dollar pricing succeeds, it would positively anchor global AI valuations; if the IPO is discounted or postponed, the narrative of an AI bubble will face interim pressure.

Follow-up Tracking

• SpaceX IPO (next month): Will it pull capital away from the public market, forcing OpenAI to adjust its roadshow timeline?
• Anthropic Q2 performance: 10.9 billion dollars in revenue + profit, if realized, will create direct narrative suppression before OpenAI's roadshow.
• Anthropic IPO timing (as early as October): Which company goes public first and who prices higher will be the most significant pricing event in the AI sector this year.
• S-1 public version: Loss amounts, Microsoft contract terms, and breakdown of the 280 billion dollar forecast assumptions; any single number exceeding expectations will trigger market repricing.
• Will Ultraman change his timing judgment: He said that market conditions are the ultimate variable.

Ultraman brought ChatGPT into the world, launching this AI revolution. But revolutions do not wait for anyone to get ready.

Now, he is sitting on the back of a tiger that has already started running, with Anthropic, SpaceX, and Google in front of him, while behind him is a seemingly bottomless bill for computing power.

Getting down is difficult. Not getting down is also hard.