OpenAI completes its largest funding round in history; Anthropic is even more eager to go public.

Author: Lin Wanwan

On March 31, 2026, OpenAI announced the completion of a $122 billion funding round, valuing the company at $852 billion, the largest private equity funding round in human business history.

Amazon invested $50 billion in OpenAI. $15 billion was immediately received, while the remaining $35 billion was contingent on a certain condition being met.

This condition is that OpenAI completes an IPO or achieves AGI.

One ​​is going public, the other is creating general intelligence that surpasses human capabilities.

The world's largest e-commerce company has staked more money than most countries' annual military spending on a single "or" situation. Let's break down OpenAI's funding rounds to see the structure. Nvidia invested $30 billion, and OpenAI happens to be one of Nvidia's largest GPU customers. OpenAI's CFO, Sarah Friar, has stated that most of the money will go back to Nvidia. Amazon invested $50 billion, and OpenAI ran its models on AWS for inference, increasing AWS revenue and improving Amazon's financial statements. Microsoft has invested over $13 billion, and OpenAI has pledged to purchase $250 billion worth of cloud services on Azure. The money circulates in a closed loop and then comes back. Wall Street calls this circular financing. Bernstein analyst Stacy Rasgon says that every such transaction deepens market concerns about circular financing. Statistics from the CFA Institute are even more unsettling, with the total amount of cross-investment and cross-purchase commitments in the AI ​​field approaching $1 trillion. The topic of revolving financing has been discussed for a whole year, and everything that needs to be said has been said. The real focus of this 122 billion yuan financing isn't how the funds circulate. It's on a more direct question: what exactly is this money buying? What is the 852 billion yuan buying? The answer is, buying time. More precisely, buying time before the IPO. OpenAI currently has monthly revenue of 2 billion yuan, annualized at approximately 24 billion US dollars. The 852 billion yuan valuation corresponds to a price-to-sales ratio of approximately 35. This multiple means the market is paying for OpenAI three or four years from now. To get a sense of this, consider these benchmarks: Nvidia's PS ratio was around 20x when it was making huge profits. Snowflake peaked at 100x but quickly fell below 30. Salesforce was around 10x when it went public. 35x is already quite aggressive for a company that's still losing money. OpenAI's own plan is to reach $100 billion in revenue and $14 billion in profit by 2029. To grow from $24 billion to $100 billion, the compound annual growth rate needs to exceed 40% for four consecutive years. I seriously considered historically any software company that has maintained this growth rate on a base of tens of billions of dollars in revenue, and I couldn't find a single one. The $852 billion valuation can only be justified under one condition: someone must be willing to buy it at that price on the open market. In other words, the IPO must succeed. Once this is understood, the entire financing structure becomes clear. Of Amazon's $50 billion, $35 billion is tied to the IPO; the money won't arrive until the listing. SoftBank's $30 billion is disbursed in three installments: the first when the financing closes, and the subsequent two in July and October, precisely timed to coincide with key junctures in the IPO preparation period. OpenAI sold 3 billion shares to retail investors through banks for the first time and is also planning to enter ARK Invest's ETF. Retail investors bought shares and invested in ETFs, becoming the natural buying base when the IPO opens. The wording in the financing announcements no longer sounds like a report to private equity investors. "We are the fastest platform to reach 10 million users, the fastest to reach 100 million users, and soon the fastest to reach 1 billion weekly active users." "Our revenue growth rate is four times that of Google and Meta during the same period." This kind of rhetoric can be directly copied onto the first page of the prospectus without any changes. A PitchBook study points out that among the three largest AI IPO candidates—OpenAI, Anthropic, and Databricks—OpenAI has the lowest score in terms of business quality fundamentals, but the highest valuation. Every design detail of the $122 billion financing points in the same direction. Let this company go public, let the public market absorb this valuation. OpenAI needs an IPO, but it's not the only one. This is the real drama of 2026. Let's look at the queue. CoreWeave went public last March at $40, now at $130, with a market capitalization exceeding $46 billion, setting an example for other companies. Databricks is valued at $134 billion and is on its roadshow, with annualized revenue of nearly $5 billion. Cerebras resolved its CFIUS review and resubmitted its IPO application. The real heavyweights are Anthropic and OpenAI. Anthropic, valued at $380 billion, has already hired Wilson Sonsini for its IPO legal preparations. Kalshi predicts a 72% probability that Anthropic will go public before OpenAI. This odds are detrimental to OpenAI. The pool of funds available to buy AI companies is limited; if Anthropic absorbs this capital and attention first, OpenAI's IPO pricing will be squeezed. And Anthropic is indeed encroaching on OpenAI's territory. In the enterprise API market share, OpenAI's share dropped from 50% in 2023 to 25% by mid-2025, while Anthropic's rose from 12% to 32% during the same period. Anthropic's revenue growth rate is approximately three times that of OpenAI. Some analysts extrapolate from the current curve, predicting that Anthropic will surpass OpenAI's annualized revenue by mid-2026. Two years ago, OpenAI dominated the enterprise market; now, Anthropic is the leader in the enterprise API market. Claude Code alone generates $2.5 billion in annualized revenue and contributes 4% of global GitHub public commits. This rapid turnaround is rare in the tech industry. OpenAI certainly has its own trump cards. With 900 million weekly active users, 50 million paid subscribers, and an annualized revenue exceeding 100 million RMB after six years of advertising trials, ChatGPT's brand awareness and user habits remain its biggest competitive advantage in the AI ​​industry. However, the slowdown on the enterprise side is a real issue. Both companies are also spending money at an alarming rate. OpenAI is projected to lose 14 billion RMB in 2026, with an annualized cash burn rate potentially reaching 57 billion RMB by 2027. The 122 billion RMB funding round sounds astronomical, but it will likely only sustain the company for about 18 to 24 months. Anthropic is projected to spend 19 billion RMB in 2026, with 12 billion RMB for model training and 7 billion RMB for inference. Whoever goes public first will have a better chance of survival. The private equity market is already running out of money to feed these companies; the public market is the last tap that hasn't been turned on yet. Renaissance Capital predicts that there could be 200 to 230 IPOs in 2026, and the combined IPO fundraising of just four companies—OpenAI, Anthropic, Databricks, and Cerebras—could exceed $200 billion. This is the largest tech IPO window since 2000. The last time there was an IPO wave of this magnitude was also in 2000. Can the speed of making money outpace the speed of spending money? All valuations, all financing structures, and all IPO plans ultimately boil down to one judgment: AI can make money faster than it spends. If they outperform, the $122 billion funding was visionary, and the $852 billion valuation is a discount. However, some are already modeling scenarios where they can't outperform. Analysts call it the CapEx Cliff. With hundreds of billions of dollars in data centers built, the revenue from the software running on them won't cover costs. An efficiency revolution will replace the scale race, and companies that bet everything on "bigger is better" will find themselves sitting on expensive but underutilized hardware. Efficiency advancements are happening faster than most people realize. Training a model equivalent to GPT-4 cost approximately $79 million in 2023; by 2026, with next-generation hardware and technologies like distillation and quantization, the cost has dropped to $5 million to $10 million. Last year, DeepSeek R1 trained a near-state-of-the-art inference model with less than $300,000. This January, it published a new training architecture paper, continuing its focus on efficiency. Google's latest Gemini 3.1 Flash-Lite has reduced the inference price to $0.25 per million tokens. IBM researchers have publicly stated that 2026 will be the year that the paths of cutting-edge large models and efficient small models diverge. If the efficiency route continues to outperform the scale route, OpenAI's computing empire, built with $852 billion in funding and a valuation of $100 billion, may face devaluation before it's even fully completed. After the bursting of the bubble economy in 2000, the internet didn't disappear; Google rose from the ruins. Those companies that died were those that raised the most money and built the most infrastructure at the peak of the bubble, but never found a sustainable business model. AI won't disappear either. But whether the $122 billion valuation and $852 billion can sustain them until profitability is far from certain. The drum is still beating, and the beat is getting faster and faster.