OpenAI and peers fork out millions as they compete for top researchers

Silicon Valley’s leading AI labs – OpenAI, Google DeepMind, and Elon Musk’s xAI, are reportedly in a tough contest for top researchers as the battle for AI supremacy rages on.

The competition is also coming at a price – high compensation packages, way beyond the usual salaries in the tech industry, according to a Reuters report. And investors also seem attracted to a team of experts.

OpenAI’s ChatGPT changed a lot of things

Since the debut of ChatGPT in late 2022, firms have intensified recruiting to “superstar” levels, with veteran insiders likening the scramble to landing top-tier athletes.

Ariel Herbert-Voss, former OpenAI researcher and now CEO of cybersecurity startup RunSybil, describes these hires as chess pieces. Labs map out candidate expertise, “Do I have enough rooks? Enough knights?” and pay premium sums to fill gaps quickly, he says.

Speed in assembling the right mix of specialists can make or break the next breakthrough.

These so-called “ICs” (individual contributors) are prized for their ability to push frontier research in large-language models and other cutting-edge domains. Noam Brown, whose work on advanced reasoning powered OpenAI’s recent AI wins, recalls a whirlwind job search in 2023.

He was invited to lunch with Sergey Brin, joined poker nights at Sam Altman’s, and even toured a private jet arranged by eager investors. Musk himself reportedly called top prospects to recruit for xAI.

Yet Brown ultimately chose OpenAI, not because it offered the richest package, but because it pledged unparalleled resources, both human and computational, to back his research.

“It wasn’t the most lucrative,” he notes, “but it was the most aligned with what I wanted to achieve.” According to Reuters, seven sources confirm that firms routinely dangle multi-million-dollar bonuses and equity to secure or retain star talent.

For example, a handful of OpenAI researchers interested in following former chief scientist Ilya Sutskever to his startup, SSI, were offered $2 million in retention bonuses plus equity uplifts of $20 million or more, contingent on just a one-year tenure.

Others fielding overtures from Eleven Labs received at least $1 million to stay put. Top AI researchers at OpenAI now report annual compensation exceeding $10 million.

Senior engineers boast hefty packages

Google DeepMind has matched and even surpassed these figures: off-cycle equity awards tailored to AI talent, vesting schedules accelerated from four to three years, and total packages reaching $20 million per annum.

Through contrast, senior engineers at major tech firms average $281,000 in salary and $261,000 in equity annually, according to industry tracker Comprehensive.io.

What heightens this arms race is the very small size of the “superstar” pool. Sources estimate just a few dozen to around a thousand researchers worldwide whose contributions are deemed pivotal to AI model development.

OpenAI CEO Sam Altman captured this notion in a late 2023 tweet, observing that while ten times better software engineers (“10x”) have long been celebrated, today’s top AI researchers are “10,000x” more impactful.

The departure of OpenAI’s CTO Mira Murati in September further intensified the fight. After founding her own startup and recruiting some 20 former colleagues, Murati has since expanded her team to roughly 60, even before releasing a product. Investors are reportedly lining up for what could become a record-setting seed round, banking on the team’s collective expertise.

Scarcity has driven companies to innovate in scouting talent. Zeki Data, a firm specializing in AI recruitment, applies analytics methods borrowed from sports, à la “Moneyball,” to uncover promising researchers overlooked by standard pipelines.

Its analyses reveal, for instance, that Anthropic is targeting experts in theoretical physics while other labs recruit quantum computing specialists.

“Today’s fast progress is attracting brilliant minds from diverse fields,” says Sébastien Bubeck, who left Microsoft last year to join OpenAI as VP of GenAI research. “I have mathematicians on my team who would not have entered AI if not for these rapid advances. Their insights are already making a tangible difference.”