OpenAI's Betrayal: How an Open Company Became Closed

The Nonprofit That Became AI's Most Powerful Corporation

When Elon Musk helped co-found OpenAI in 2015, his stated mission was singular: prevent artificial general intelligence from falling into the hands of corporate giants like Google. A decade later, his lawyers stand in federal court trying to stop OpenAI itself — the original counterweight — from becoming the very thing it was built to counterbalance.

This transformation from idealistic nonprofit to $300 billion for-profit juggernaut represents one of the most dramatic corporate metamorphoses in Silicon Valley history. It raises fundamental questions about whether mission-driven organizations can survive contact with the economics of cutting-edge AI development.

Key Takeaways

• OpenAI was founded in 2015 as a nonprofit dedicated to ensuring AGI benefits 'all of humanity'
• The organization has undergone multiple structural changes, culminating in a planned conversion to a for-profit benefit corporation
• Elon Musk's lawsuit alleges OpenAI abandoned its founding mission and original agreements
• The company's valuation has soared to approximately $300 billion, making it one of the most valuable private companies globally
• Core researchers and co-founders, including Ilya Sutskever, have departed amid internal disagreements over the company's direction
• The case illuminates a broader tension in AI: can open, safety-first development survive the industry's capital demands?

The Promise That Started It All

Silicon Valley has never been short on idealism baked into corporate charters. But OpenAI's promise was different in both scale and ambition. The organization didn't claim it would build better search, faster ad systems, or cheaper cloud services. It declared it would develop artificial general intelligence and ensure that technology benefited all of humanity.

In 2015, that language didn't sound absurd. ChatGPT didn't exist yet. GPUs hadn't become the new gold. Sam Altman wasn't yet the most prominent figure in the AI era. The founding team — which included Musk, Altman, Greg Brockman, Ilya Sutskever, and others — pledged $1 billion to an open research lab that would publish its findings freely and serve as a counterweight to Google's growing dominance in AI through its DeepMind acquisition.

The nonprofit structure wasn't an accident. It was the architecture of a promise. By operating without shareholders or profit motives, OpenAI signaled that its loyalty belonged to humanity, not investors. Research papers were published openly. Code was shared on GitHub. The organization operated with a transparency that matched its name.

The Economics of AI Forced a Reckoning

The first cracks appeared around 2018 and 2019, when the cost of training large language models began to escalate dramatically. GPT-2, released in 2019, was already pushing the boundaries of what a nonprofit budget could sustain. The team recognized a brutal economic reality: building AGI — or anything approaching it — required billions of dollars in compute infrastructure.

In 2019, OpenAI created a 'capped-profit' subsidiary, a novel legal structure that allowed outside investment while theoretically maintaining nonprofit oversight. The cap limited investor returns to 100x their investment — a figure that sounds restrictive until you consider that early investors in a $300 billion company would find that cap remarkably generous.

Microsoft became the primary beneficiary of this new structure, investing an initial $1 billion in 2019 and eventually committing over $13 billion in total. In exchange, Microsoft received exclusive cloud computing rights, API access, and integration privileges that transformed products like Bing, Copilot, and Azure.

Critics argue this was the moment OpenAI effectively became a subsidiary of one of the world's largest technology corporations — precisely the outcome its founding charter was designed to prevent.

• 2015: Founded as a pure nonprofit with $1 billion in pledged donations
• 2019: Created capped-profit subsidiary; Microsoft invests $1 billion
• 2023: Microsoft's total investment reaches $13 billion; ChatGPT launches to 100 million users
• 2024: Valuation reaches $157 billion after funding round
• 2025: Planned conversion to for-profit benefit corporation; valuation hits $300 billion

Musk's Lawsuit Exposes the Fault Lines

Elon Musk's legal challenge against OpenAI, filed initially in early 2024 and later refiled with expanded claims, alleges that Altman and other leaders engaged in a 'bait and switch' of extraordinary proportions. The suit claims Musk was induced to contribute time, money, and credibility based on promises of openness and nonprofit governance that were systematically dismantled.

Musk departed OpenAI's board in 2018, reportedly after a failed attempt to take control of the organization himself. His motivations are complicated — he went on to found xAI and build Grok, making him a direct competitor. But the legal arguments his team raises resonate beyond personal grievances.

The core question is deceptively simple: can an organization solicit donations and talent under a nonprofit mission, then convert that accumulated value into a for-profit enterprise? OpenAI's planned transition to a benefit corporation structure would effectively complete this transformation, removing the nonprofit board's ultimate authority over the company's direction.

A federal judge allowed key parts of Musk's lawsuit to proceed, signaling that the legal questions are substantive, regardless of the plaintiff's mixed motives.

The Exodus of True Believers

Perhaps more telling than any lawsuit is the departure of key figures who built OpenAI's technical foundation. Ilya Sutskever, the co-founder and chief scientist who was instrumental in the dramatic boardroom crisis of November 2023 — when Altman was briefly fired — left to start Safe Superintelligence Inc., a company focused exclusively on AI safety.

Other notable departures include:

• Dario Amodei and Daniela Amodei, who left in 2021 to found Anthropic, now valued at over $60 billion and backed by Amazon and Google
• Andrej Karpathy, a founding member who departed twice, citing a desire to work on different projects
• Jan Leike, who co-led OpenAI's superalignment team before resigning and publicly stating that 'safety culture and processes have taken a back seat to shiny products'
• Mira Murati, former CTO, who departed in September 2024 after 6 years at the company

These aren't disgruntled middle managers. They are the architects of the technology that made OpenAI valuable. Their collective departure suggests that the tension between OpenAI's original mission and its commercial trajectory isn't merely a legal dispute — it's a philosophical schism that runs through the organization's core.

What 'Open' Means When Nothing Is Open

The irony embedded in OpenAI's name has become increasingly difficult to ignore. The company that once published research freely and released model weights for public use now guards its technology behind proprietary APIs and commercial licensing agreements.

GPT-4's technical report, released in March 2023, was notable for what it didn't contain: virtually no details about architecture, training data, or compute resources. The company cited safety and competitive concerns. Critics saw a company using 'safety' as a convenient shield for standard corporate secrecy.

Compare this to Meta's approach with Llama 3, which released model weights openly, or Mistral's commitment to open-source releases. Even Google, the original boogeyman that OpenAI was created to counter, has released open models through its Gemma family.

The argument that safety requires secrecy is not without merit. But it becomes harder to sustain when the same company simultaneously pursues a $300 billion valuation and partners exclusively with the world's second-most-valuable corporation. Safety and commercial interest align conveniently when both point toward keeping technology proprietary.

The Broader Industry Implications

OpenAI's transformation matters beyond its own corporate drama because it tests a fundamental thesis about AI development. Can the technology that may prove most consequential in human history be developed responsibly within conventional corporate structures?

The evidence so far is not encouraging. Every major AI lab faces similar tensions:

• Google DeepMind merged its safety-focused research lab into a product-driven corporate structure
• Anthropic accepts billions from Amazon while positioning itself as the 'responsible' alternative
• Meta open-sources models partly to undermine competitors' business models, not purely from altruism
• xAI, Musk's own venture, has raised over $6 billion with no pretense of nonprofit governance

The market has effectively decided that AGI development is a corporate activity. The question is whether any governance structure — nonprofit, benefit corporation, or otherwise — can meaningfully constrain commercial incentives once the technology proves valuable enough.

Looking Ahead: What Comes Next

OpenAI's planned conversion to a for-profit benefit corporation is expected to complete in 2025, pending legal challenges and regulatory approval. Several state attorneys general have expressed interest in reviewing the conversion, and Musk's lawsuit adds another layer of legal uncertainty.

The outcome will set precedents far beyond one company. If OpenAI successfully converts, it validates a playbook: use nonprofit status to attract talent, donations, and public trust during the risky early stages, then convert to for-profit once the technology proves commercially viable. Every future AI nonprofit will operate under the shadow of this precedent.

Sam Altman has argued that the benefit corporation structure actually provides more durable protection for OpenAI's mission than the current hybrid model. Whether you find that argument persuasive likely depends on how much faith you place in corporate governance documents versus the gravitational pull of $300 billion in enterprise value.

What remains undeniable is this: the organization founded to ensure AI remained open, safe, and beneficial to all of humanity has become a closed, commercially aggressive entity valued at more than Goldman Sachs or Boeing. The counterweight has become the weight. And the debate over whether that transformation was inevitable — or a choice made by specific people at specific moments — will shape how we govern AI development for decades to come.

The founding promise of OpenAI was always ambitious to the point of audacity. The question history will judge is whether it was ever sincere.