The $110 billion financing is actually a huge gamble. OpenAI must achieve AGI or conduct an IPO this year.

$110 billion. This is the largest single - round financing I've seen since entering the media industry, and it's also the largest single - round financing for an unlisted company in human history. OpenAI achieved this feat at the end of February 2026.

To understand how exaggerated this figure is, let's make a few comparisons.

The financing amount of Saudi Aramco, the world's largest IPO, was $29.4 billion, which is less than one - third of OpenAI's current round of financing.

In this round of financing, Amazon invested $50 billion, while NVIDIA and SoftBank each invested $30 billion.

However, according to foreign media reports, $35 billion of Amazon's investment is conditional. The funds will only be transferred if OpenAI achieves AGI or completes an IPO by the end of the year.

Before the investment, OpenAI's valuation was $730 billion, and after the investment, it reached $840 billion. For comparison, JPMorgan Chase has a market value of $830 billion, and SpaceX is valued at $800 billion.

But what I want to express is that behind this financing lies a grander narrative.

Tech giants are using real money to buy a ticket, attempting to take control of the planet's AI future.

OpenAI, which has secured the highest single - round financing in history, is just one of the chips in this high - stakes gamble.

Why is OpenAI worth so much?

Currently, OpenAI is in a tight spot. On one hand, there is a record - breaking financing figure; on the other hand, there are ever - expanding losses.

In 2025, OpenAI's revenue reached $13.1 billion, exceeding the original target of $10 billion. This figure looks quite impressive, exceeding expectations by more than 30%. However, when placed in the overall financial statement, the situation is not so optimistic.

In the same year, the company lost $8 billion. In 2026, this loss is expected to soar to $25 billion. According to OpenAI's internal forecast, from now until 2029, the cumulative loss will reach $115 billion, and OpenAI may not turn a profit until the 2030s.

In 2026, OpenAI's burn rate is expected to remain at 83.3%, which is rare in the history of startups.

However, the increase in burning money is only one aspect. The key is that OpenAI is losing its market share.

ChatGPT was once the absolute leader in the AI chatbot market, with a market share of up to 86.7% at the beginning of 2025. But by the beginning of 2026, this figure had dropped to around 64.5%. With the rise of Anthropic and Google, ChatGPT's market share may continue to decline.

So, the question is, since OpenAI is in such a difficult situation, why are these tech giants still willing to invest $110 billion?

The answer lies in the list of investors.

Amazon, NVIDIA, and SoftBank. Their investments are far more than just simple financial support. NVIDIA is both an investor in OpenAI and a GPU supplier. Amazon and Microsoft not only invest in OpenAI but also sell cloud computing services to it. SoftBank is OpenAI's partner in the Stargate Project.

This means that a significant portion of the financing will actually be returned to the investors in the form of services, forming a closed loop of funds and resources. OpenAI needs NVIDIA's GPUs to train models, Amazon and Microsoft's cloud services to run inferences, and SoftBank's funds to build data centers.

These giants can not only profit from OpenAI's growth through investment but also lock in long - term business contracts.

Moreover, this is also a high - stakes gamble for future dominance.

These tech giants want to pour money into OpenAI while it is still unlisted to grab more shares and influence. Once OpenAI successfully completes an IPO in the second half of 2026, these early investments will translate into substantial control over OpenAI.

They are betting not only on OpenAI as a company but also on the future lifestyle based on AI or AGI. This is why Amazon's bet agreement includes AGI and listing.

Whoever controls OpenAI may define the next - generation technical standards, business rules, and user experience. The astronomical figure of $110 billion is the price of the entrance ticket to this game.

OpenAI itself is also aware that this amount of money is far from enough.

The company disclosed that it will invest approximately $600 billion in computing power by 2030. A few months ago, Altman also said that they would invest $1.4 trillion in infrastructure.

The specific flow of these funds includes large - scale data center construction, GPU procurement, talent competition, and R & D investment.

This scale of expenditure has gone beyond ordinary business logic. It is more like an infrastructure race, similar to the railway construction or power grid laying in the past. This time, the competition is about computing power and AI capabilities.

OpenAI's valuation logic is based on the premise that whoever builds a large - scale AI infrastructure first and forms a computing power hegemony will dominate the future.

From this perspective, the $110 - billion financing and the post - investment valuation of $840 billion no longer seem so outrageous.

This is not about valuing a loss - making startup but about pricing an infrastructure platform that may reshape the entire technology industry.

However, the question is whether this premise can hold, which is still full of uncertainties.

The Stargate Project is dragging OpenAI down

If financing is the life - support for OpenAI, then the Stargate Project is a constantly bleeding wound.

In January 2025, US President Trump announced the Stargate Project at the White House. OpenAI, Oracle, and SoftBank will jointly invest a total of $500 billion over four years to build AI data centers across the United States.

This project was undoubtedly a milestone for the US AI industry at that time, promising to create more than 100,000 jobs.

However, more than a year after the press conference, the joint - venture company of the Stargate Project has not recruited any employees and has not actually developed any data centers. Only employees from OpenAI, SoftBank, and Oracle are temporarily working on the project.

The three - party partners are at an impasse on basic issues such as responsibility division, site ownership, and fund flow, and the project progress is slow.

The core reason for the stagnation of the Stargate Project is the disagreement among the three parties on control. SoftBank hopes to take the lead in the project and own the site. OpenAI wants to directly control the data centers to ensure the stability of computing power supply. Oracle, as the infrastructure provider, also has its own interests.

The fact is that one monk fetches water to drink, two monks carry water to drink, and three monks have no water to drink.

From September to October 2025, over more than a month, OpenAI executives flew to Tokyo multiple times to negotiate with Masayoshi Son, but they made little progress.

The project delay caused OpenAI to miss the critical construction window and lose several general contractors, forcing it to find new partners.

Desperate, OpenAI once tried to build its own data centers bypassing its partners.

But lenders were reluctant to provide funds to OpenAI. The reason is simple. Although we see OpenAI as a leading global technology company, in the eyes of banks, it is just a company with an unproven business model, burning billions of dollars every year, with a name that contains "Open" (open - source) but whose models and technologies are hardly open - source. Capital doesn't believe that OpenAI can repay the money.

After hitting a wall, OpenAI had to return to the negotiation table.

Previously, OpenAI bypassed the three - party joint - venture company of the Stargate Project and directly reached a bilateral agreement with Oracle to build 4.5 - gigawatt data centers in multiple locations in the United States, excluding SoftBank.

According to the agreement terms, if there are delays or cost overruns, both parties will share the additional costs. If there are any savings, both parties will also benefit.

Subsequently, Altman announced with high - profile that OpenAI would add five new data centers, and he also said that the total computing power of these data centers would be 7.5 gigawatts.

Unfortunately, due to the delay of the Stargate Project, the construction plans for these five data centers were all put on hold.

Ultimately, OpenAI failed to achieve the goal of signing contracts for 10 - gigawatt computing power through Oracle and SoftBank by the end of 2025.

OpenAI has reached an agreement for a capacity of approximately 6.8 gigawatts, but there are no actually built data centers, only the planned capacity in the contract.

To fill the gap, OpenAI had to purchase temporary computing power from competitors such as AWS and Google Cloud.

This not only increased costs but also made OpenAI more strategically dependent on these cloud service providers. Originally, OpenAI wanted to get rid of its dependence on cloud service providers by building its own data centers. However, due to the failure of the Stargate Project, it has fallen into a deeper dependence.

The Stargate Project has exposed OpenAI's predicament.

The company needs a large amount of computing power to train and run models, and building its own data centers is the only way to reduce costs in the long run. However, building its own data centers requires huge upfront investment, and OpenAI's current financial situation cannot support such investment.

This creates a vicious cycle: Without enough computing power, it cannot maintain its technological leadership; to obtain computing power, it must rely on external suppliers or partners; relying on external suppliers will increase costs and reduce profit margins; low profit margins make it even more difficult to obtain loans to build its own infrastructure.

The Stargate Project was originally an opportunity to break this cycle, but the competition for control among the three parties has turned this opportunity into a bubble. Now, OpenAI can only continue to struggle in this cycle, using the financing to buy expensive computing power, then continuing to lose money, and then continuing to seek financing.

This also explains why OpenAI needs $110 - billion in financing. This money is not only used to cover daily operating costs but also to purchase computing power, pay for cloud services, and prepare for future data center construction.

However, even with this money, if the Stargate Project continues to stagnate, OpenAI's computing power dilemma cannot be fundamentally solved.

Anthropic and OpenAI start the IPO race

Two weeks before OpenAI announced its financing, its biggest competitor, Anthropic, completed a $30 - billion Series G financing, with a post - investment valuation of $380 billion, more than double its valuation in September 2025.

This round of financing was led by Singapore's sovereign wealth fund GIC and hedge fund Coatue, with tech giants such as Microsoft, NVIDIA, Amazon, and Google following suit.

Interestingly, except for SoftBank, the players investing in OpenAI are also investing in Anthropic. These tech giants are placing bets on both companies to ensure that they won't miss out no matter who ultimately wins.

Anthropic's growth trajectory is in sharp contrast to OpenAI's. The company's annualized revenue has reached $14 billion, growing more than tenfold each year in the past three years. Among them, the annualized revenue of Claude Code has exceeded $2.5 billion, and its revenue has more than doubled since the beginning of 2026. The number of enterprise subscriptions has increased fourfold.

This gives Anthropic the confidence to achieve profitability faster.

Anthropic expects to break even in 2028, two years earlier than OpenAI's target in 2030. At the same time, Anthropic also said that its burn rate in 2026 will drop to about one - third of its revenue and further drop to 9% in 2027, far lower than OpenAI's.

Behind this difference are the completely different strategic choices of the two companies.

OpenAI focuses on the consumer market. ChatGPT has more than 900 million weekly active users and 50 million paying subscribers. Anthropic focuses on the enterprise market, with 80% of its revenue coming from enterprise customers. Thanks to Claude Code's excellent performance in code generation, Claude is even more popular among programmers than ChatGPT.

According to the prediction of data analysis agency Epoch AI, if both companies maintain their current growth trends, Anthropic may overtake OpenAI to become the number one in terms of revenue in the AI field between 2026 and 2027. At that time, the annualized revenues of the two companies will meet at around $43 billion.

Although Epoch AI mentioned a detail, that is, Anthropic's growth rate has slowed down from tenfold per year to sevenfold per year since July 2025, they believe that the stability and willingness to pay in the enterprise - level market give investors more confidence in Anthropic's long - term business situation.

Both companies are also preparing for an IPO in the second half of 2026. This will surely be the largest IPO in the history of technology, without a doubt.

OpenAI has had informal contacts with Wall Street investment banks and has hired Chief Accounting Officer Ajmere Dale and Head of Investor Relations Cynthia Gaylor. Anthropic has hired Silicon Valley law firm Wilson Sonsini to start preparations for the IPO.

Interestingly, people are not very optimistic about these two companies.

The prediction market agency Polymarket estimates that the probability of OpenAI going public in 2026 is only 51.5%.

Furthermore, Polymarket also points out that the probability of Anthropic going public before OpenAI is only 28%. In other words, neither Anthropic nor OpenAI is likely to go public this year.

Previously, Altman once said that he was not in a hurry to go public. However, after Anthropic announced its listing plan, OpenAI immediately accelerated the entire IPO process.

Because if Anthropic goes public first, it will gain greater flexibility in the capital market and can continue to raise funds through the public market. Meanwhile, OpenAI is still stuck in the private market, having to renegotiate each round of financing, diluting its equity and accepting various conditions from investors.

The gap will gradually widen.

The deeper problem is that both companies are betting on the same assumption, that is, the scaling law is still valid, and investing more computing power and data can lead to stronger capabilities.

However, once this assumption fails and the improvement of model capabilities reaches a bottleneck, the AI bubble will burst, and these astronomical valuations will collapse instantly.