What makes people say ‘no’ to a $100 million job offer? The AI talent war reveals lessons for hanging on to top performers 

Why the AI talent race is happening

For money or the mission

Culture’s quiet power 

The team wins

What to do to keep talented employees

Lila MacLellan is a senior

Even if you don’t know the name Andrew Tulloch, you’ve probably heard something about his story. 

Tulloch is one of Silicon Valley’s cutting-edge AI researchers who have lately attracted astonishingly lucrative job offers from legacy tech companies trying to lure them away from their AI-native startups. In Tulloch’s case, the legacy player was Meta, which reportedly offered the young researcher $1.5 billion over at least six years to leave his current post at Mira Murati’s Thinking Machines Lab. Meta has disputed this figure, but other big numbers have been confirmed

In many instances, the fat figures worked: Individual targets have been poached and entire startups have been purchased. But what’s been more surprising are the reports that when Meta CEO Mark Zuckerberg and other tech royalty knocked on some doors, including Tulloch’s, AI researchers sometimes said, “No, thanks.” CEOs of newer companies and established firms—Dario Amodei at Anthropic, Sam Altman at OpenAI, and Lisa Su at AMD—have even said they will not match poaching offers to keep their talented employees. Just as strikingly, some said that they haven’t needed to do so to keep their top performers. 

(Fortune reached out to Tulloch and several others who reportedly turned down huge pay packages, and none agreed to talk. Meta recently announced a hiring freeze in its artificial intelligence division.)

So do we need to rethink the power of cash in negotiations? Is the era of the superstar employee over?

Not exactly, say management experts. Many have been watching the AI talent race and considering what it means for CEOs of companies in other industries and of more modest means. But they do believe that while competitive compensation still matters, the AI talent wars prove that squishy concepts like culture, empathetic leadership, and collegiality are stronger forces than most people would imagine. 

Ajay Agrawal, professor of entrepreneurship at the University of Toronto’s Rotman School of Management and co-

If the AI industry evolves the way other recent disruptive technologies have, Agrawal explains, one foundational large language model will take the greatest market share in the near future, the way Google came to dominate search engines, even though other options existed in Google’s early days. So, all the companies spending billions on AI also hope to build the “winning” model to amortize their costs across hundreds of millions of users. “That just warrants big investments in general, whether it’s for equipment or people,” the professor says.

But unlike other new technologies, AI models are also trained

Knowing this, says Agrawal, “Nobody wants to fool around with a second-rate team.”

The promise of great wealth also explains why some AI researchers have chosen to stay in their current jobs. In conversations with some of the bright young minds who have been caught up in the talent wars, Agrawal, who is also a research fellow at Stanford’s Digital Economy Lab, has learned that many are gambling on a payoff in the future. “Even though the offers sound very big to you and me,” Agrawal says, “they are with a younger, smaller company where they have equity, and they think there’s a chance that their equity in that company will be worth more.” That tantalizing possibility just isn’t as likely at an established brand.

Still others who have rejected fat offers claim they’re less motivated

Indeed, several CEOs have positioned their company’s mission as their competitive moat in the talent wars, one that will appeal to altruistic types in an age when smart young people are being called on to cultivate their “moral ambition.” In a statement emailed to Fortune, for example, Anthropic said: “Top AI talent choose Anthropic because they want to build AI systems for society’s benefit with safety principles prioritized from day one. We attract and retain talent because of our research quality, our commitment to AI safety, and our track record of industry-shifting breakthroughs.” The company also touted its top leaders as talent magnets.  

Relatedly, as Agrawal notes, larger players like Google, Meta, and Microsoft have all existed long enough to have accumulated scandals and ethical crises attached to their names—even if they too started life with mission statements tied to making the world a better place. Newer companies promising noble causes, in contrast, are not tarnished—or less tarnished, at least—which also appeals to idealistic young workers. AI-native companies “don’t have the baggage,” the professor says, “so they can create missions and visions with a blank canvas.”

That last point brings us to the realm of corporate culture, the least visible of forces at play in these scenarios, but one that some academics say is the ultimate filter that employees use to make decisions. 

Culture—the way organizations behave as opposed to the goals they set for the organization—can dictate who is attracted to a company, who stays, and who leaves. Culture is what people are referring to when they describe Mira Murati, who left OpenAI as its chief technology officer to start Thinking Machines, and had 20 employees follow her, as a low-ego leader. “At OpenAI, she was known for her emotional intelligence and lack of ego, which earned her the loyalty of the research and engineering

Jennifer Chatman, dean of the University of California, Berkeley’s Haas School of Business, and a scholar who has studied organizational culture for decades, tells Fortune that companies in every industry need to pay attention to salaries in the marketplace because culture will never compensate for less robust pay. But keeping pay competitive, she says, “gives people an opportunity to think hard about how they would fit with the orientation of the organization.” That’s a good thing, she adds. “When people actively select-in based on the culture, there are all kinds of good outcomes that emerge: They perform better, they stay longer, they’re more committed. They can move through the organization more effectively.”

In some company cultures, recruiting high performers

Similarly, CEOs tend to fall into one of two camps: They elevate star players at all costs, often reinforcing an idea that a single, usually young and male, genius can make or break an organization, or they build strong teams and support a collective intelligence. Research shows that the latter is far more sustainable, says Haas. Under a star model, “you have people that can really dial up their capability and do more for the organization, and you can be lean,” Chatman says, “but when they leave, it’s a big deal.” 

Increasingly, the team model seems to be preferred

Peter Schein, cofounder of OCLI.org, a consulting company he launched with his father, the late MIT scholar Edgar Schein, endorses these views. The most effective leaders cultivate a culture of learning and information sharing across teams, he says. “The brilliant individual who has a brilliant vision for how to take [your model] to the next release, those people are always going to be around,” Schein says. “But I will also bet on the team nine out of 10 times over the individual.”Schein’s advice for leaders who want to hang on to their people, or become the kind of CEO to whom talented employees may one day work with again, is simple:  “You have an easier time retaining people who you’ve actually built a rapport with, not treating people with this professional distance that says these are [just] human re

“These are humans who happen to be a re