Brockman testifies Musk pushed to commercialize OpenAI

The OpenAI trial is turning into a fight over what the company was always supposed to become. On Tuesday, Greg Brockman told the court that Elon Musk wasn’t some pure nonprofit idealist who later got betrayed. Brockman’s version is almost the opposite — Musk supported a for-profit structure back in 2017, but wanted control of it. The reason matters because Musk is suing OpenAI over its shift away from the original nonprofit model, and this testimony goes straight at the core of that claim. ### What did Brockman actually say? Brockman testified in federal court in Oakland on May 5 that Musk pushed to reorganize OpenAI as a for-profit company during internal discussions in 2017. But Brockman said Musk also wanted majority equity, board control, and the CEO job. Brockman framed that as a takeover attempt, not a principled objection to commercialization. Where does Mars come into this? This is the weird-sounding detail, but it’s central. Brockman said Musk talked about needing around $80 billion to build a city on Mars and saw OpenAI as something that could help generate that kind of capital. In other words, the testimony paints Musk as someone who viewed commercialization as useful — just useful on terms he controlled. ### Why is that such a big deal legally? Because Musk’s case depends on the idea that OpenAI broke faith with its founding mission. He has argued that the organization was supposed to develop AI for humanity’s benefit, not turn into a profit engine tied to Microsoft and giant private financing rounds. If the jury believes Brockman, then Musk was not opposing commercialization itself. He was opposing a version where he didn’t run it. ### Why does compute spending keep coming up? Brockman also told the court that OpenAI expects to spend about $50 billion on computing power in 2026. He said compute costs were about $30 million in 2017 and have since exploded into the tens of billions. That number is doing a lot of work in the case. It helps OpenAI argue that frontier AI stopped being a lab-scale science project years ago and became an industrial capital problem. ### So was OpenAI’s structure always going to break? Basically, that is OpenAI’s argument. The company is trying to show that the original nonprofit shell made sense when the bills were small and the mission was mostly research. But once model training started demanding massive clusters, energy, and chips, the old structure. ### What else came out in court? The trial has also surfaced how personal and messy the split became. Brockman described Musk as helpful in some recruiting conversations but polarizing in others, and other coverage from the courtroom focused on a tense 2017 meeting over ownership and control. That doesn’t decide the legal question by itself, but it helps explain why the breakup hardened into a years-long feud. ### What is the real argument underneath all this? It’s about whether AI labs can stay mission-driven once the cost of staying at the frontier hits nation-state scale. OpenAI is saying the money requirements forced evolution. Musk is saying the evolution violated the original deal. Brockman’s testimony matters because it tries to prove that even Musk saw the commercial logic early — he just wanted the keys. ### Bottom line? This was more than courtroom gossip about Mars. Brockman gave the jury a simple story: Musk wanted OpenAI to become commercial too, and the real fight was over control. If that story lands, Musk’s case gets a lot narrower — from “they changed the mission” to “they changed it without me.”