In an effort to regain lost ground in the race for dominance in artificial intelligence, Facebook founder Mark Zuckerberg reached out several months ago to Mira Murati, the former Chief Technology Officer of OpenAI, proposing to buy her new startup, Thinking Machines Lab.
When she refused, Meta’s CEO intensified his efforts. Over the following weeks, he contacted more than a dozen employees at Murati’s company—which has around 50 staff members—attempting to convince them to leave her. His primary target was Andrew Tallak, a prominent researcher and co-founder of Thinking Machines.
To entice him, Zuckerberg reportedly offered a compensation package—including bonuses and stock options—that could total up to $1.5 billion over at least six years. For perspective, Zuckerberg’s offer to Mira Murati herself was $1 billion.
Tallak declined. And none of Murati’s colleagues left.
A Meta spokesperson, Andy Stone, dismissed the reports as inaccurate and “ridiculous,” adding that any compensation depends on stock performance. He also stated that Meta has no interest in acquiring Thinking Machines.
Loyalty Beyond Money
Even by Silicon Valley standards, rejecting nine-figure offers is rare. But as competition for AI talent intensifies, it’s clear that even the richest companies cannot always win.
Many researchers operate as independent contractors, but many choose to remain loyal to their leaders—almost mythical figures in tech circles. The unique cultures of various startups reinforce this commitment. At the same time, companies are increasingly working to defend against ongoing “talent poaching.”
OpenAI and companies founded by its former leaders—including Murati’s—have been continuous targets of Zuckerberg’s recruitment efforts. Since the early stages of the AI race, pioneers in this field have pledged to participate in a historic mission: creating Artificial General Intelligence (AGI)—a system smarter than humans at almost all tasks.
OpenAI, founded by Sam Altman, Elon Musk, Ilya Sutskever, and others, has framed this mission almost religiously, positioning itself as a nonprofit focused on “benefiting humanity.”
Competition with OpenAI, Anthropic, and SSI
Meta has already approached over 100 OpenAI employees, hiring at least ten. On July 25, Zuckerberg appointed researcher Shengjia Zhao, who has three years of experience at OpenAI, to lead Meta’s new “superintelligence” team.
OpenAI employees who refused to leave cited their belief that OpenAI is closer to achieving AGI, their preference for smaller teams, and their desire not to have their achievements exploited primarily for ad revenue.
Zuckerberg has also courted Anthropic, a $170 billion startup founded by Dario Amodei, a former OpenAI executive who left five years ago with top scientists. All seven Anthropic co-founders remain at the company and are connected through the “effective altruism” movement, which early on expressed concerns about unchecked AI risks. Some even lived in a cooperative in San Francisco.
From Anthropic, Meta hired two former directors—Joel Pobara and Anton Bakhtin—who previously worked at Meta. Ilya Sutskever, co-founder of Safe Superintelligence (SSI), built his company to make it extremely difficult for employees to be poached. Unlike Amodei, he has preferred to train new, unknown technologists with fresh ideas rather than bring over many ex-OpenAI collaborators.
Employees avoid mentioning SSI on LinkedIn profiles to prevent aggressive recruitment. It is said Zuckerberg even proposed buying SSI, but the offer was refused.
Who is Mira Murati and Her Loyal Team
Murati, who spent six years at OpenAI before leaving in September, gathered around her an “apostles’ team.” Of Albanian origin, she helped launch OpenAI’s first product and eventually led nearly every department as CTO. She is known for her emotional intelligence and lack of ego—qualities that strengthened her team’s loyalty.
At Thinking Machines, she implemented a similar “flat” organizational structure where senior researchers are simply “technical staff,” reminiscent of the legendary Bell Labs.
When she founded the company in February last year, over 20 former OpenAI colleagues joined, including co-founder John Schulman, one of the architects of ChatGPT, who previously worked at Anthropic. Many of Murati’s researchers come from OpenAI’s “post-training” group, responsible for training models to interact with humans.
The mission of Thinking Machines remains mostly confidential, even to some recent investors who contributed $2 billion. According to Murati, the company is building “multimodal AI that interacts naturally with the world” and is ready to launch its first product in the coming months. The company is headquartered in San Francisco’s Mission district, just blocks away from OpenAI offices.
Before founding Thinking Machines, Murati was a key figure at OpenAI, the company behind groundbreaking AI technologies like ChatGPT, DALL·E, and Codex. Her contributions significantly advanced AI’s accessibility in industries such as education, healthcare, design, and programming.
Murati was not only a technical leader but also a strategic visionary. She emphasized aligning AI development with safety and responsibility, pushing OpenAI to consider social implications of powerful AI systems.
Often called the “brain behind AI,” Murati is praised by colleagues for her calm determination and clear thinking. She combined mechanical precision with a human-centered approach and paved the way for her new venture.
In 2025, she spoke at the World Economic Forum in Davos, stating, “Artificial Intelligence without values is intelligence without conscience.” This sparked international debate on AI ethics, and she now advises the European Commission on new AI regulations—an extraordinary role for a startup founder.
Murati’s leadership is internationally recognized. She was named among the 100 most influential people in AI (2024) by Time magazine and one of the 100 most powerful women in business (2023) by Fortune. With her growing team and rejection of billion-dollar offers from Meta, Murati has secured her place as a key shaper of AI’s future.
Who is Andrew Tallak?
Tallak, a co-founder of Thinking Machines, was a central figure in Zuckerberg’s and Meta’s new superintelligence lab head Alexander Wang’s ongoing recruitment efforts. Both bombarded him with messages.
Originally from Australia, Tallak graduated from the University of Sydney with top honors in science. He worked 18 months in Facebook’s machine learning department before moving to Cambridge for postgraduate studies and then returning to California to join Facebook AI Research. He rose to become a distinguished engineer, one of the highest technical positions in the company.
In 2016, OpenAI president Greg Brockman tried to recruit him as one of their earliest employees, telling Elon Musk in an email that Tallak earned $800,000 a year at Facebook and would likely seek a raise. At that time, OpenAI offered $175,000 base salary and $125,000 bonuses.
“Andrew is very close to saying yes, but worried about the pay cut,” Brockman wrote on February 21. Tallak did not join OpenAI then, but he eventually joined seven years later when ChatGPT had taken the world by storm and the startup’s valuation had skyrocketed.
