840 billion, the largest financing in history is here

Ask AI · How can Amazon’s investment conditions accelerate OpenAI’s IPO process?

Just as the market is buzzing about OpenAI’s imminent trillion-dollar IPO, the company has dropped another heavy bomb.

On March 31, OpenAI officially announced that it had completed a financing deal totaling up to $122 billion (approximately 8.4 trillion yuan at current exchange rates), with the company’s post-investment valuation soaring to $852 billion (about 5.9 trillion yuan).

This figure not only broke OpenAI’s own funding record but also set the largest single private funding record in the global private market ever.

OpenAI’s Chief Financial Officer Sarah Friar also commented: “This financing even makes the largest IPO in history look modest.” She stated that amid multiple uncertainties in the public markets, this funding will provide the company with ample operational flexibility to ensure steady investment in computing resources and AI technology R&D.

With the closing of this round of financing, market expectations for OpenAI’s IPO continue to rise. It is widely anticipated that the company will initiate its listing process as early as this year.

Largest Ever Funding: Led by Three Giants, Retail Investors Enter for the First Time

The preparation for this round of financing took several months, and its scale and structure are unprecedented. The total of $122 billion surpasses the $110 billion committed in February this year and makes any previous tech funding rounds pale in comparison.

The core funds for this round mainly come from three leading tech giants: Amazon pledged to invest $50 billion, Nvidia and SoftBank Group each committed $30 billion. These commitments of $110 billion form the foundation of this financing.

Notably, Amazon’s $50 billion investment has clear preconditions: $35 billion of this must be invested contingent upon OpenAI completing an IPO or reaching a general artificial intelligence milestone. This conditional clause tightly links Amazon’s interests with OpenAI’s IPO process, adding substantial pressure to the latter’s listing expectations.

Nvidia and SoftBank’s respective commitments of $30 billion are paid in installments, with two payments of $10 billion each scheduled for July 1 and October 1 this year.

As a long-term partner of OpenAI, Microsoft also participated in this round, though its specific investment amount has not been disclosed. By the end of 2025, Microsoft’s cumulative investment in OpenAI has exceeded $13 billion.

In addition to equity financing, OpenAI announced it would expand its revolving credit line to about $4.7 billion, supported by top global banks including JPMorgan Chase, Citibank, Goldman Sachs, and Morgan Stanley. The company stated that this credit line has not yet been drawn upon and aims to enhance financial flexibility and reserve capacity for large-scale future capital expenditures.

Another major feature of this round is its openness to individual investors. For the first time, OpenAI raised over $3 billion from high-net-worth individuals through banking channels, further diversifying its funding sources.

Meanwhile, OpenAI will be included in several exchange-traded funds managed by Ark Invest under Cathie Wood, providing an indirect channel for ordinary investors to hold stakes in the company. This arrangement broadens the shareholder structure ahead of the anticipated IPO.

Beyond these core investors, the funding round attracted dozens of top global institutions, including Andreessen Horowitz (a16z), Abu Dhabi Sovereign Fund MGX, D.E. Shaw Ventures, TPG, T. Rowe Price, Altimeter, Appaloosa, BlackRock, Blackstone, Coatue, Sequoia Capital, Fidelity, Temasek, and others, making the investor lineup highly elite.

This record-breaking financing by OpenAI signifies more than just a company milestone. In an official statement, OpenAI said: “Such moments are rare. In past generations, capital markets helped build systems that define the modern economy—from electricity to highways to the internet. Now, it’s happening again. The capital deployed today is helping to build the infrastructure of intelligence itself.”

This capital will be used to build “intelligent infrastructure,” which the company believes will, over time, generate value across the economy, enterprises, communities, and individuals.

$2 Billion Monthly Revenue, Still Burning Cash

Alongside the financing announcement, OpenAI unusually disclosed detailed financial data, revealing the true scale of this AI giant.

OpenAI disclosed that its current monthly revenue is about $2 billion, with annual revenue for 2024 reaching $13.1 billion. The company claims its revenue growth rate is four times that of giants like Google and Meta, which define the internet and mobile eras.

In terms of revenue structure, enterprise-level business is growing rapidly and now accounts for 40% of total revenue, expected to rise to 50% by the end of 2026. This indicates OpenAI is shifting from a consumer AI company toward deep penetration into enterprise services.

According to The Information, OpenAI has recently raised its revenue forecast for the next five years—aiming for up to $284 billion (about 1.96 trillion yuan) by 2030.

OpenAI also revealed an astonishing user base. As of March this year, ChatGPT’s weekly active users exceeded 900 million, with paid subscriptions surpassing 50 million. OpenAI claims that ChatGPT’s monthly web visits and mobile session volumes are six times those of the second-ranked AI app, and the total time users spend on AI is four times that of all other AI applications combined.

Recently, OpenAI’s advertising pilot project achieved over $100 million in annual recurring revenue within less than six weeks, opening a highly promising new revenue channel. This also indirectly indicates that as AI agents become more widespread, the model of “cheap, casual use” AI is approaching its limit, and strategic resource allocation will become critical.

Despite impressive revenue figures, OpenAI is still in a burn phase and has not yet turned a profit. In 2024, the company’s annual expenditure is about $8 billion, mainly for AI chips procurement, data center expansion, and recruiting top talent at high salaries.

Pressure is ever-present. Analysts point out that how to justify the $852 billion valuation to public market investors amid ongoing losses will be a core challenge for CEO Sam Altman. In recent months, OpenAI has cut some large expenditure plans and shut down certain features and products to control costs. The company is working to optimize operations and pave the way for its upcoming IPO.

Sora Abandoned, Fully Betting on “Super Apps”

One of the biggest moves in OpenAI’s recent strategic adjustments is the shutdown of Sora.

Sora was an AI-generated video model developed by OpenAI, capable of creating realistic videos up to 60 seconds long based on text prompts. It could generate complex scenes with multiple characters and specific movements, deeply simulating real physics. Launched in February 2024, it was downloaded over 1 million times within five days.

Sora was officially shut down in March 2026. Its lifecycle was only two years, but it experienced the full arc from “technological shock” to “commercial failure.”

According to multiple media reports, the main reason for shutting down Sora was high operating costs and severe losses. Forbes estimated that Sora’s annual operating costs reached over $840B, burning about $15 million daily. In contrast, Sora’s total global revenue since launch was only about $1.4 million to $2.1 million, a stark difference from ChatGPT’s $1.9 billion revenue during the same period.

“It was a very difficult decision, but it all comes down to compute power,” explained an OpenAI executive regarding the decision to exit the video AI service Sora.

Video generation consumes significantly more computing resources than text or image generation. OpenAI’s decision to cease Sora’s service aims to concentrate valuable compute resources on developing the new foundational model “Spud” (internal code name, meaning “potato”) for ChatGPT.

This decision also reveals a deeper reality in the AI industry: the supply shortage of high-performance semiconductors and rising electricity prices have pushed the endless consumption of compute power to its limits. When compute becomes the scarcest resource, OpenAI must make tough choices.

Strategically, OpenAI clearly states that “AI super apps” are its next core goal. The company plans to integrate ChatGPT, Codex, browsing capabilities, and other intelligent agents into a single platform to create a unified system capable of understanding user intent and executing tasks across applications.

The company believes that as model capabilities continue to improve, the bottleneck limiting AI adoption has shifted from intelligence itself to usability. Users need not multiple scattered tools but a unified intelligent interface.

Looking further ahead, OpenAI’s Chief Scientist Jakub Pachocki recently revealed that the company is targeting an unprecedented research goal: to create an “AI researcher” capable of autonomously solving complex problems by 2028.

This is a fully automated multi-agent research system that can independently handle scientific tasks across mathematics, physics, chemistry, biology, and policy analysis. In September this year, the first phase will be launched—an “autonomous AI research intern.” Pachocki said achieving the “AI researcher” plan is OpenAI’s “North Star” goal for the coming years.

In January, OpenAI released Codex, an intelligent agent capable of instant code generation and executing complex calculations. It can analyze documents, generate charts, organize emails, and summarize social media. Today, Codex has become a standard tool for internal staff, assisting in coding and problem-solving. Pachocki said Codex can be seen as a prototype of the “AI researcher,” with future innovations expected to be disruptive.

Additionally, it was announced that on February 15, 2026, Peter Steinberger, the developer of “Lobster,” officially joined OpenAI to lead Codex development.