Elon Musk Wanted to Control OpenAI. Here's Why the Company Said No.

OpenAI has released internal messages showing that Elon Musk asked for majority ownership, complete control, and the CEO job at a proposed for-profit company in fall 2017. OpenAI rejected the terms, saying they conflicted with the organization's core mission. The newly published messages also provide details about Musk's early involvement with OpenAI and why he eventually stepped back.

According to the published communications, the conflict came down to a simple problem: OpenAI was running out of money. In early 2017, as the organization's research became more ambitious, leaders realized they would need billions of dollars to buy and run the powerful computers needed to develop artificial general intelligence — an amount far beyond what nonprofit funding could provide.

What Musk Proposed

Through 2017, Musk became more involved in OpenAI's plans. In July, he discussed merging OpenAI with a hardware company — a signal that he was thinking about reshaping how the organization worked.

By September 2017, Musk had started a company called "Open Artificial Intelligence Technologies, Inc." That same month, he presented his terms for converting OpenAI into a profit-making company: he would own the majority of shares, have complete decision-making power, and be the CEO.

OpenAI's leadership said no in September 2017. They explained that letting one person have total control would go against why OpenAI was founded in the first place. This decision meant Musk would no longer shape OpenAI's future, though he stayed on the board until February 2018.

The Money Problem

Tensions continued into 2018. In December, Musk told OpenAI's leaders they needed to raise "billions per year immediately or forget it." By that time, OpenAI was already searching for other ways to fund its work.

The money issue was very real. Training the kind of artificial intelligence systems OpenAI was building requires massive computer power. Think of it like running thousands of very powerful machines all working together on the same task for months on end. The costs had grown far beyond what anyone expected when OpenAI started.

Why This Matters

Disputes over control and money between founders have happened before in the AI world. When an organization doing research hits the point where it needs huge amounts of capital, something has to give. In some cases, a larger company buys the research organization outright — Google bought DeepMind in 2014, for example. In other cases, the organization finds a different way forward.

Over my thirty years covering technology, I have watched this pattern repeat in different industries. When the cost of development gets very high, research-focused organizations face hard choices. The semiconductor industry went through the same squeeze in the 1980s and 1990s, when building factories to make computer chips became so expensive that most independent labs could not afford to do it anymore.

One thing worth noting: OpenAI's leadership recognized years before the rest of the industry that building advanced AI would require enormous computing resources. That insight turned out to be correct, and it shaped every major decision the company has made since.

The Hardware Question

The proposed merger with a hardware company is interesting because it points to a strategy that worked well for other tech companies. Google and Microsoft today both design their own AI systems and also control the computer infrastructure those systems run on. Owning both sides of that equation gives them advantages in speed and cost.

If Musk had gotten his way and taken control of OpenAI, the company would have looked very different. It might have become part of his other businesses — his electric car company Tesla or his space company SpaceX. Instead, OpenAI chose a different path: it partnered with Microsoft for computer resources while keeping its research focused broadly on AI, not on any single company's needs.

The timing of these negotiations is worth remembering because transformer architectures — a type of AI system that turned out to be crucial — were just beginning to show promise in 2017. The massive computer power that Musk and OpenAI's leaders were discussing would indeed become necessary as those systems grew more advanced. It just took a few more years for the rest of the world to catch up and understand why.

What Changed Because of This

The newly published messages show us some of the key decisions that shaped today's AI industry. Had Musk's plan succeeded, OpenAI would have developed in a very different direction, probably tied closely to his other companies and their goals.

What actually happened instead was that OpenAI found an alternative. The company partnered with Microsoft for funding and computing resources, and that partnership helped create the AI systems — especially the GPT models — that have become central to how many people interact with artificial intelligence today.

These messages reveal that fundamental questions about how AI research should be governed, how it should be funded, and whether a mission-driven approach can survive in a capital-intensive world were being debated at the highest levels long before the general public even knew these questions mattered. The way OpenAI resolved this particular conflict with Musk established a model that continues to influence how AI companies are built and run today.