Court filings just revealed that Microsoft nearly walked away from OpenAI before writing the checks that made ChatGPT possible.
The Summary
- Emails from 2018 show Microsoft executives were deeply skeptical of OpenAI, but feared Amazon would scoop them up if they passed.
- Musk's lawsuit against Altman is surfacing internal documents that question whether any CEO can be trusted with superintelligence development.
- The evidence reveals the power dynamics that shaped the AI industry years before GPT-4 existed.
The Signal
Microsoft's leadership didn't believe in OpenAI. They believed in denying OpenAI to competitors. The 2018 emails surfacing in Musk's lawsuit show executives viewing the nonprofit as a risky bet, but one they couldn't afford to let Amazon place first. This wasn't visionary investment. It was defensive positioning.
The timing matters. In 2018, OpenAI was still a nonprofit research lab burning cash with no product. GPT-2 wouldn't drop until February 2019. Microsoft was being asked to fund what looked like an expensive science project with unclear returns.
"Microsoft feared pushing OpenAI into the arms of Amazon more than they trusted OpenAI's mission."
But here's what the emails reveal about how the agent economy actually gets built: not through bold conviction in a technological future, but through cold calculation about who controls the infrastructure. Microsoft wasn't buying OpenAI's vision. They were buying Amazon's inability to buy it.
The lawsuit itself raises a harder question: can Sam Altman, or any single CEO, be trusted to develop superintelligence? Musk's legal team is using OpenAI's own safety promises against it. Every blog post about responsible AI development. Every commitment to keeping AGI beneficial. All of it is now evidence of promises broken.
Key tensions the emails expose:
- Microsoft's skepticism vs. their fear of being left out
- OpenAI's nonprofit mission vs. its need for corporate capital
- Public safety commitments vs. private competitive pressure
The discovery process is doing what journalism couldn't: showing us the actual decision-making behind the companies building Web4. Not the press releases. Not the mission statements. The spreadsheets and the risk assessments and the "what if Amazon gets there first" late-night emails.
The Implication
Every major AI lab is now one lawsuit away from having their internal doubts become public record. If you're building agents, tokenizing models, or promising your investors the future, remember: your emails are forever, and discovery is undefeated.
Watch for more documents. The Musk lawsuit is a discovery goldmine, and we're just seeing the first samples. The people who wrote checks for OpenAI in 2018 had doubts. Find out what the people writing checks in 2026 are saying in private.