OpenAI completes the largest funding round in history, while Anthropic is more eager to go public than it is.

By | Lin Wanwan

On March 31, 2026, OpenAI announced the completion of a $122 billion funding round, with a $852 billion valuation— the largest private financing in human business history.

Amazon invested $50 billion in OpenAI. Of that, $15 billion was transferred immediately, while the remaining $35 billion would be paid only after a certain condition was met.

That condition is for OpenAI to complete an IPO, or to achieve AGI.

One is going public, the other is building general intelligence that surpasses humans. The biggest e-commerce company on Earth is betting a sum of money—higher than the annual military spending of most countries—on an “or.”

Let’s break down all of OpenAI’s funding and look at its structure.

NVIDIA put in $30 billion, and OpenAI just happens to be one of NVIDIA’s largest GPU customers.

OpenAI’s CFO Sarah Friar has said herself that most of the money would go back to NVIDIA.

Amazon’s $50 billion investment comes in; OpenAI runs inference by hosting the models on AWS; AWS revenue rises, and Amazon’s financial reports look better. Microsoft has invested more than $13 billion cumulatively, and OpenAI has committed to purchasing $250 billion worth of cloud services on Azure.

The money went around in a closed loop and came back. Wall Street calls this circular financing—looped financing.

Bernstein analyst Stacy Rasgon said that every one of these transactions deepens market concerns about circular financing. The statistics from the CFA Institute are even more unsettling: the total value of commitments in the AI space involving all kinds of cross-investing and cross-procuring already sits at nearly $1 trillion.

But this topic of circular financing has been discussed for a whole year. Everything that needed to be said has already been said.

The $122 billion round that truly deserves attention isn’t about how the funds circle. It’s about a more direct question: what exactly is all this money buying?

What $852 billion buys

The answer is: time. More precisely, time up until before the IPO.

OpenAI’s current monthly revenue is $2 billion, or about $24 billion on an annualized basis. A $852 billion valuation corresponds to roughly 35x sales. This multiple implies that the market is paying for OpenAI three to four years in the future.

Let’s get a feel by looking at a few benchmarks. When NVIDIA is making money like crazy, its PS is roughly 20x. Snowflake reached as high as 100x at its peak, but quickly fell back below 30x. Back when Salesforce went public, it was around 10x.

35x on a company that’s still losing money is already aggressive.

OpenAI’s own plan is $100 billion in revenue by 2029 and $14 billion in profit. To go from $24 billion to $100 billion, the annual compound growth rate needs to stay above 40% continuously for four years. I seriously thought about it—of software companies that maintained that kind of growth on a revenue base in the hundreds of billions— I couldn’t find even one.

For a $852 billion valuation to hold, there’s only one condition: someone is willing to take over that valuation at that price in the public market. In other words, the IPO must succeed.

Once you see that, the entire financing structure becomes easy to read.

Within Amazon’s $50 billion, $35 billion is tied to the IPO condition—if it doesn’t go public, the money won’t be paid. SoftBank’s $30 billion is split into three tranches. When the financing closes, the first payment is made; the subsequent two payments land in July and October, precisely placed on the key timing nodes of the IPO preparation period.

For OpenAI, the first time it sold $3 billion worth of shares to retail investors via banks, it also had to enter ARK Invest’s ETF. Retail investors buy the shares and then get into the ETF. When the IPO opens, it’s naturally a foundation of buy-side demand.

The wording in the funding announcement also no longer sounds much like a report to private equity investors. “We’re 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 at the same time.” You can paste this set of talking points directly into page one of the prospectus without changing it.

PitchBook has a research note pointing out that among the three largest AI IPO candidates—OpenAI, Anthropic, and Databricks—OpenAI’s business quality fundamental score is basically the lowest, but its valuation is the highest.

Every design detail of the $122 billion round points in the same direction: get this company to go public, and have the public market take that valuation.

Two companies race for the same water faucet

OpenAI needs an IPO, but it isn’t the only one that does. This is the real big drama of 2026.

First, look at the queue. CoreWeave already went public last March. Its offer price was $40; now it’s $130, market cap over $46 billion—setting an example for the companies after it. Databricks has a $134 billion valuation and during roadshows it reported annualized revenue nearing $5 billion. Cerebras resolved the CFIUS review by resubmitting an IPO application.

The real heavyweights are Anthropic and OpenAI. Anthropic’s valuation is $380 billion, and it has already hired Wilson Sonsini to prepare for IPO legal work. Kalshi’s prediction market estimates the probability that Anthropic will list before OpenAI at 72%.

These odds are brutal for OpenAI. The pool of capital seeking to buy AI assets is limited. If Anthropic eats up that pool of funding and attention first, OpenAI’s IPO pricing will be compressed.

And Anthropic is indeed encroaching on OpenAI’s turf. In the enterprise API market share, OpenAI fell from 50% in 2023 to 25% by mid-2025; over the same period, Anthropic rose from 12% to 32%. Anthropic’s revenue growth rate is about three times OpenAI’s. Some analysts, extrapolating from the current trajectory, expect Anthropic to surpass OpenAI’s annualized revenue in mid-2026.

Two years ago, OpenAI dominated the enterprise market. Now Anthropic is already the leader in the enterprise API market. Claude Code’s annualized revenue from a single product is $2.5 billion, contributing 4% of global GitHub public submissions. This speed of reversal is rarely seen in the tech industry.

OpenAI certainly has its own ace. With 900 million weekly active users, 50 million paid subscriptions, and an advertising business that tested the waters for six years and already surpassed $100 million in revenue, ChatGPT’s brand awareness and user habits are still the biggest moat in the AI industry. But the slowdown on the enterprise side is real and tangible.

The two companies are also spending money at astonishing speed at the same time.

OpenAI is expected to lose $14 billion in 2026; by 2027, its annualized burn rate could be as high as $57 billion. The $122 billion funding sounds astronomical, and it could support them for roughly 18 to 24 months. Anthropic is expected to spend $19 billion in 2026: $12 billion training models and $7 billion running inference.

Whoever goes public first gets to survive longer. Private equity money is already running out of feed for these companies; the public market is the last faucet that hasn’t been turned on. Renaissance Capital predicts there could be 200 to 230 IPOs in 2026. Just summing OpenAI, Anthropic, Databricks, and Cerebras, the total IPO funding size could exceed $200 billion.

This is the biggest tech IPO window since 2000. And the last time IPOs at this scale appeared as a wave was also in 2000.

Can the pace of making money outrun the pace of spending

All valuations, all financing structures, all IPO plans ultimately hinge on one judgment: that AI’s speed of making money can beat the speed of spending.

If it does, then the $122 billion funding round is foresight, and the $852 billion valuation is a discounted price.

There are also models being built for scenarios where it doesn’t.

Analysts call it the CapEx Cliff—the capital expenditure cliff. A few trillion dollars’ worth of data centers get built. The money made by the software running on them isn’t enough to cover costs. Then an efficiency revolution replaces the scale race. Companies that bet everything on “the bigger, the better” find themselves sitting on piles of expensive hardware with underutilization.

Improvements in efficiency are faster than most people realize. Training a model at roughly the same level as GPT-4 cost about $79 million in 2023. By 2026, using next-generation hardware plus techniques like distillation and quantization, costs have already dropped to between $5 million and $10 million.

Last year, DeepSeek R1 trained an inference model close to the cutting edge with less than $300,000. This year, in January, it published a new training architecture paper again, continuing to work on efficiency. Google’s latest Gemini 3.1 Flash-Lite has pushed inference pricing down to $0.25 per million tokens. Researchers at IBM said publicly that 2026 will be the year where the frontier large-model route and the efficient small-model route diverge.

If the efficiency route keeps outpacing the scale route, the compute empire OpenAI builds with a $852 billion valuation might face devaluation before it’s even finished being built.

After the dot-com bubble burst in 2000, the internet didn’t disappear. Google grew out of the ruins. What died were the companies that raised the most money and built the most infrastructure at the bubble’s highest point—yet never managed to touch a sustainable business model.

AI won’t disappear either. But whether a $12.2 billion round and a $852 billion valuation can last until the day they reach profitability is far from as cut-and-dried as it looks.

The drums are still being beaten, and the drumbeats are getting faster.

Click to learn more about Lydon BlockBeats’ job openings

Welcome to join Lydon BlockBeats’ official community:

Telegram subscription group: https://t.me/theblockbeats

Telegram discussion group: https://t.me/BlockBeats_App

Twitter official account: https://twitter.com/BlockBeatsAsia