Musk tried to lure Altman to Tesla to build an AI lab

Leaked messages between Shivon Zilis, president of Neuralink, and Tesla executives reveal a long-standing, ambitious plan by Elon Musk: to build his own AI lab and recruit the world's best scientists to it. The letters date from 2017, when Musk was only beginning to grasp the scale of the changes under way in artificial intelligence.

Why Musk wanted his own lab

By 2017, Musk was already disillusioned with OpenAI. The organization, which he had helped found in 2015 as a nonprofit, was drifting further and further from his vision. It had once been seen as a counterweight to Google DeepMind, but its pace of development seemed too slow to Musk. Meanwhile, Musk was getting more deeply involved in Tesla's business and starting to see enormous potential in AI for autonomous cars.

Musk came to a simple conclusion: if he wanted to control the development of AI, he needed his own lab. Tesla had three main advantages: financial resources, unique data from millions of drivers around the world, and boundless ambition.

The leaked letters show that Musk was not just dreaming in public. He was seriously working out a concrete plan to bring top scientists into the company and even sketched out candidates.

Whom Musk wanted to poach

There were two main candidates. The first was Sam Altman, then working at OpenAI. Altman had already made a name as an experienced manager and strategist, having built his reputation at Y Combinator. The second was Demis Hassabis, co-founder of DeepMind and creator of AlphaGo, the system that in 2016 beat the world Go champion time after time, stunning the world.

Both were at the absolute peak of their careers and had deep knowledge of how to build large-scale AI systems. Landing either of them would have given Tesla a huge strategic advantage:

• Hands-on experience building and scaling large AI research systems
• Personal ties to the world's best researchers and the ability to recruit them
• A reputation that automatically attracts funding and PR
• A deep understanding of how to build long-term research organizations
• The ability to compete directly with OpenAI and DeepMind on their own turf

Why it didn't work out

In practice, none of Musk's plans came to fruition. Altman stayed at OpenAI and later became its CEO, a position he holds to this day. Hassabis chose DeepMind and Google, where he continues to lead the company. Musk did not get his man. His attempt to seize AI leadership through Tesla failed.

But that did not stop him. Musk kept believing in the need for his own AI project and returned to the idea again and again. Eventually, years later, in 2023, he launched xAI, a separate company devoted entirely to developing artificial intelligence. The Tesla episode shows how Musk thinks: he can wait, he can re-plan, but he never gives up on his ideas.

What this means

The disclosure of these letters completely overturns the narrative of a "chance" rivalry between Musk and the leaders of the AI industry. It turns out this is a conflict with deep historical roots, one that did not begin yesterday, over loud tweets, or over personal grudges. Behind the public statements and sharp clashes lie real strategic calculations, long-held ambitions and fundamental disagreements over how artificial intelligence should develop.