The AI team Zuckerberg can't buy

Former OpenAI CTO Mira Murati wants to help companies build their own AI models with Thinking Machines Lab – with top talent that even Zuckerberg would love to have.

The AI battle between the tech giants is reaching a new level: Meta CEO Mark Zuckerberg attempted to acquire the AI startup of former OpenAI CTO Mira Murati . After Murati rejected his takeover offer, Zuckerberg apparently wanted to lure some of the brightest minds from her team at Thinking Machines Lab (TML) – and he didn't skimp on the money.

According to a Wall Street Journal report, Zuckerberg wanted to bring the AI elite on board at any cost. The mega-offer was targeted at researchers from Murati's new team. Twelve people from the 50-person team were approached. All of them reportedly rejected Meta's lucrative offers, The Economic Times quoted Murati as saying in a statement to Wired.

Australian AI expert Andre Tulloch was even reportedly offered a compensation package of $1 billion (€867 million) over six years—consisting of stocks, bonuses, and salary. An offer that is absurdly high, even in hypercompetitive Silicon Valley.

Since the launch of ChatGPT and the growing influence of OpenAI, Meta has felt increasingly under pressure in the AI race. While the company was able to generate attention with Llama, Meta's open-source approach to generative AI, it now appears to be resorting to more drastic measures in the race for top talent.

Mira Murati, who left OpenAI in early 2024, is considered one of the key minds behind GPT-4. Her new company, Thinking Machines, is currently still in stealth mode. But the mere fact that Meta was willing to invest so much money in her team speaks volumes about its strategic importance.

Despite the gigantic offer, Murati's colleagues decided not to follow Meta's call. The reasons for this remain largely speculation. But industry insiders suspect that it's about more than money. Many researchers in the AI scene now value independent research, ethical principles, and greater creative freedom – something that's harder to achieve in a large corporation like Meta.

In addition, Thinking Machines appears to have an ambitious vision that goes beyond what Meta can currently offer.

Murati was able to bring high-profile former OpenAI employees on board. OpenAI co-founder John Schulman, former Head of Special Projects Jonathan Lachman, and former VPs Barret Zoph and Lilian Wenig joined Thinking Machines Lab.

Former OpenAI Chief Research Officer Bob McGrew, OpenAI computer scientist Alex Radford and Alexander Kirillov, who previously worked with Murati on ChatGPT's voice mode, are also now said to be working at TML, as Techcrunch reported.

But top scientists don't work for Murati purely out of interest in research. According to US H-1B visa data , which US companies must submit when hiring non-US citizens, Thinking Machines Lab pays technical staff an annual salary of between $450,000 and $500,000 (€380,860–€423,200).

The average base salary for techies is $462,500, excluding bonuses or compensation.

In comparison, the average base salary for technical staff at OpenAI is around $292,000 (€247,240), according to the documents. The highest-paying position there is $530,000 (€448,570), and the lowest is $200,000 (€169,270).

At Anthropic, technical employees earn an average of about $387,500 (€327,950), with salaries ranging from $300,000 (€253,900) to $690,000 (€583,990).

What exactly the startup does remains unknown. There is no product yet. The website simply states that TML aims to bridge the gap between rapidly growing AI capabilities and the actual understanding of the technology—with a focus on transparency, open science, and open-source code. The startup plans to regularly publish research results, technical blog posts, and code to engage the entire AI community and accelerate research.

In June, Murati's vision convinced the first investors. A seed round raised $2 billion (€1.7 billion), valuing the company at over $12 billion (€10 billion). The round was led by a16z. Nvidia, Accel, ServiceNow, Cisco, AMD, and Jane Street also participated.

This is one of the largest seed funding rounds in Silicon Valley history, insiders told the Financial Times .