Elon,
Working towards AGI, OpenAI has a lot of issues. The current research that's coming out of OpenAI is good, but the rate of progress feels nowhere near where it would need to be to win the race. The team is doing great work but they are starved of compute and constantly distracted by the infrastructure problem.
Google has TPUs, infinite money, the best researchers, and a clear path to AGI through DeepMind plus Brain. Facebook is hiring aggressively. Microsoft, Amazon, Apple are all building internal labs.
OpenAI as a nonprofit cannot win this race. It does not have the budget to retain its top researchers; we are already losing them to other labs at a rate that is accelerating. Sam's response to this is the for-profit conversion, but a for-profit OpenAI is just a worse Google: same incentives, smaller budget, worse infrastructure.
There is one alternative that could work. Tesla is the only large company that needs AGI to deliver on its product roadmap. Self-driving is an AGI-complete problem. Optimus is an AGI-complete problem. Tesla has the data, the chips, the manufacturing, the cash flow.
OpenAI should attach to Tesla as its cash cow. It is my opinion that this is the only path that could work. Imagine Tesla as the equivalent of Google, and OpenAI as the equivalent of DeepMind, sharing the same fate.
Andrej