OpenAI raises $100 billion in funding, the largest in human history, but Open has died.

OpenAI is completing the largest private funding round in human business history: over $100 billion, with a valuation approaching $850 billion; but ten years ago, its nonprofit goal of “benefiting all humanity” has long been abandoned.
(Background: Financial Times reports NVIDIA will invest $30 billion in OpenAI, replacing last year’s $10 billion partnership deal)
(Additional context: Amazon is rumored to invest hundreds of millions in OpenAI, pushing its own chip Trainium to challenge NVIDIA’s dominance)

Table of Contents

• An Open Letter
• Powering Dreams with Compute
• Five Days and Five Years
• The $100 Billion Bill
• The Echoes of Open

On December 11, 2015, a group of top Silicon Valley AI researchers published an open letter. Announcing the formation of a nonprofit called OpenAI, they pledged to “advance digital intelligence in a way that is most likely to benefit all of humanity.” All research outputs would be open source. Profit was not the goal; safety was. Founders included Sam Altman, Elon Musk, Ilya Sutskever, Greg Brockman, with an initial commitment of $1 billion.

Ten years later, in February 2026, the organization is completing the largest private funding round in human business history: over $100 billion, with a valuation nearing $850 billion.

How big is $100 billion? This number exceeds the annual GDP of more than 140 countries worldwide. It surpasses the yearly output of mid-sized economies like Vietnam, Hungary, Morocco, and others—yet this is just a single funding round for one company. Amazon plans to invest $50 billion, SoftBank $30 billion, Nvidia $30 billion, with Microsoft co-investing; all parties expect to finalize allocations by the end of February.

This funding will surely be recorded in business history, but OpenAI is no longer a nonprofit, nor does it open-source its core models. “Open” remains in the name, but it has disappeared from the company for a long time… In this article, we’ll explore the company’s growth story.

An Open Letter

Let’s go back to 2015, when the AI industry was very different from today. In January, Google acquired DeepMind for over $500 million, sparking fears that core AI technology would be monopolized by a few tech giants. Musk and Altman shared a common concern:

If the most powerful AI systems are controlled by a single company, it’s dangerous for humanity.

So they chose a nonprofit structure. OpenAI would have no shareholders, no pursuit of profit, and no capitulation to capital. Its only obligation was to be responsible to humanity. All research would be open source, accessible for anyone to use and improve.

This choice seemed reasonable—almost noble—at the time. But it contained a fatal assumption: that the cost of AI research was manageable.

In 2015, training a cutting-edge AI model cost around several hundred thousand dollars. By 2019, with GPT-2, costs rose to the millions. In 2020, training GPT-3 was estimated between $4.6 million and $12 million. By 2023, training GPT-4 exceeded $100 million.

In plain terms: each generation’s training cost is 3 to 10 times that of the previous one. The nonprofit relied on donations and sponsorships, but the cost curve of AI research was rising far faster than any donor’s willingness or ability to fund it.

Musk sensed the problem as early as 2017. He proposed becoming CEO of OpenAI or integrating it into Tesla. Altman and Brockman refused.

In 2018, Musk stepped down from the board, citing “conflict of interest with Tesla’s AI efforts,” but the seeds of discord had already been planted.

Fast forward to 2024, Musk sues OpenAI and Altman, accusing them of “betraying the nonprofit mission.” OpenAI counters, claiming Musk supported the creation of a for-profit structure as early as 2017. The legal battle is expected to go to court in March 2026.

Ironically, the dispute itself reveals the core issue: Musk claims Altman betrayed the ideals; Altman claims Musk wanted control from the start. Regardless of who’s right, the conclusion is the same: a nonprofit organization cannot sustain the costs of AI arms races.

Powering Dreams with Compute

In March 2019, OpenAI made its most significant structural decision: establishing a “capped-profit” subsidiary.

The design was as follows: the nonprofit parent would continue to exist, but a for-profit entity would be created underneath, allowing external investors to inject capital and earn returns. However, returns were capped at 100 times the investment. Any profits beyond that would go entirely to the nonprofit parent.

The intent was to “strike a balance”: attract capital while ensuring the mission wouldn’t be swallowed by commercial interests. The nonprofit retained ultimate control; the profit entity was responsible for monetization. It seemed clever.

But once capital entered, it wouldn’t stay in the living room.

In July 2019, Microsoft became the first major investor, injecting $1 billion. By January 2023, Microsoft’s total investment reached $13 billion, giving it a 49% profit share.

In plain language: a subsidiary of a nonprofit, with nearly half its profits going to a $3 trillion market cap tech giant.

Dario Amodei saw where this was heading. As VP of research, he led GPT-2 and GPT-3 development. But he observed a troubling trend: as Microsoft’s influence grew, safety research was deprioritized. When the biggest funder demands “fast product release,” safety voices are pushed to the sidelines.

In January 2021, Amodei left OpenAI with seven core researchers to start Anthropic. That same year, OpenAI stopped open-sourcing its core models. GPT-3’s API became paid, but the model weights were no longer public.

The word “Open” in the name no longer holds true in a technical sense.

This is how the compute tyranny operates: the more successful your product, the more users you have, the higher the inference costs. Training the next generation of models requires exponentially more compute, and capital. Every new infusion of capital proportionally dilutes the original nonprofit mission.

OpenAI’s founders designed a clever structure to protect their ideals. But they didn’t foresee that AI’s cost curve would rise so steeply that no governance could keep up.

Five Days and Five Years

On a Friday, November 17, 2023, just after 1 p.m., four members of OpenAI’s board voted to oust CEO Sam Altman.

The official statement was brief: “Altman was not always fully transparent in communications with the board, impairing its ability to fulfill its responsibilities.”

Deeper issues surfaced gradually. Over the summer, a board member discovered that OpenAI’s “startup fund” was not operating as planned; an investigation revealed Altman personally held the fund, creating a serious conflict of interest under nonprofit governance.

Additionally, two senior executives provided documents describing a “toxic environment” and “trust issues.” Earlier, in November 2022, when ChatGPT launched, board members only learned about it from Twitter.

But what happened in the five days after the vote reveals more about what OpenAI has become.

Within 72 hours:

• Microsoft CEO Satya Nadella publicly supported Altman
• Over 700 OpenAI employees—almost the entire staff—signed a letter threatening mass resignation to join Microsoft
• Microsoft invited Altman to establish a new AI research division
• Investors pressured the board to reverse its decision

On November 22, Altman was reinstated. Board members Helen Toner and Tasha McCauley, who voted to oust him, were forced to leave. New board members included Bret Taylor (former Salesforce co-CEO) and Larry Summers (former U.S. Treasury Secretary).

In plain language: the nonprofit board made a decision aligned with its governance duty—to question the CEO’s integrity. But within five days, the power of capital and employees completely overturned that decision.

This is the essence of OpenAI’s identity crisis. Legally, the nonprofit board is the highest governing body, entrusted with the public mission. But in reality, the $13 billion from Microsoft and the loyalty of 700 employees are the real decisive forces.

No matter how sophisticated the governance structure, when a “nonprofit” organization’s survival depends on the attitude of a $3 trillion tech giant, “nonprofit” becomes just three words on paper.

The five days resolved the CEO issue. The next five years addressed the structural problem.

On October 28, 2025, OpenAI completed its final transformation. The nonprofit parent was restructured into the “OpenAI Foundation,” and the profit entity was officially named OpenAI Group PBC. Microsoft owns 27%, the Foundation 26%, with employees and other investors holding 47%.

Musk’s lawsuit failed to block the transition; in March 2025, a judge dismissed his injunction request.

From the 2019 “capped-profit” to the 2025 “public benefit corporation,” OpenAI took five years to complete its shift from nonprofit to for-profit. Every step was carefully framed legally to justify the change, each aimed at “raising funds for AI safety research.”

But each step moved “Open” further from its original meaning.

The $100 Billion Bill

Now, returning to the February 2026 funding round: $100 billion is not growth capital. It’s a survival bill.

In 2025, OpenAI’s annualized revenue reached $20 billion, more than doubling from $6 billion the previous year. ChatGPT’s monthly active users surpassed 300 million. By traditional software standards, this is one of the fastest revenue growth curves in history.

But OpenAI is not a traditional software company. Its cost structure is entirely different.

In 2025, OpenAI’s cloud computing expenses exceeded $8.5 billion. Add top AI researchers’ salaries (over $1 million annually), GPU procurement, data center construction, and the company burned about $17 billion in cash that year. With $20 billion in revenue, it remains deeply unprofitable.

The company’s own financial forecast is alarming: in 2026, it expects a loss of $14 billion. By 2029, cumulative losses will reach $115 billion. It will likely only break even around late 2029 or 2030.

Plainly: OpenAI must burn hundreds of millions of dollars every year for the next three to four years just to stay afloat, hoping to see profitability someday. And $100 billion is the runway it’s buying.

The investor structure of this funding itself reflects the situation:

Amazon is one of OpenAI’s cloud providers. Nvidia is its largest GPU supplier. Microsoft is both the biggest shareholder and the Azure cloud provider. As part of this partnership, OpenAI will expand its use of Amazon’s chips and cloud services.

In plain language: OpenAI’s biggest suppliers are also its biggest investors. Part of their investment will flow back to them as compute costs.

This isn’t conspiracy. It’s a unique capital cycle in the AI industry. Nvidia sells GPUs to OpenAI, profits invest back into OpenAI, which then uses the funds to buy more Nvidia GPUs. Every link is a legitimate business transaction, but together they form a self-reinforcing capital flywheel—shovels funding the gold rush.

In a recent interview, Altman admitted he has no enthusiasm for running a public company. But he also acknowledged that OpenAI’s capital needs are now so large that only the public markets can satisfy them. The company plans to file for an IPO with the SEC in late 2026, aiming for a 2027 listing, with a valuation possibly exceeding $1 trillion.

From a $1 billion donation pledge in 2015 to a $1 trillion IPO target in 2027—12 years, a 1,000-fold increase in valuation.

The Echoes of Open

OpenAI’s story is never just about funding. It’s a public experiment on whether “idealism” can survive in a capitalist world.

2015 assumption: AI is too important to be driven by profit motives.
2019 compromise: profit is allowed, but mission comes first, with profit caps.
2023 reality: capital and employees can overthrow the nonprofit board in five days.
2025 conclusion: transforming into a public benefit corporation is the only way forward.
2026 reality: $100 billion paid by suppliers and shareholders together.

Official narratives claim that the foundation structure ensures mission continuity. The foundation holds 26% of shares, can appoint the board, and commits $25 billion to health and AI resilience. The safety and security committee must include two independent directors, one of whom is a safety expert.

But the five days in November 2023 proved one thing: when legal structures clash with capital power, the legal structure loses.

Sam Altman may not be a villain; Dario Amodei may not be a traitor; Elon Musk may not be wrong. They are all struggling within an impossible equation: how to use hundreds of billions of dollars to pursue a “benefit all humanity” goal, while ensuring that money doesn’t swallow the goal itself.

The answer is embedded in OpenAI’s name. Ten years ago, it contained both method (Open) and goal (AI). Ten years later, the goal remains, but the method has already died.