OpenAI is urgently revising its compensation strategy in response to Meta’s aggressive recruitment of its senior researchers, with eight key departures reported in the past week. In a Slack memo obtained by WIRED on June 28, 2025, OpenAI’s Chief Research Officer Mark Chen expressed dismay, likening the losses to a “home invasion,” and assured employees that leadership, including CEO Sam Altman, is actively countering Meta’s offers with enhanced compensation and retention strategies.
Details of the Talent War
Meta’s Poaching Spree:
Researchers Hired: Eight OpenAI researchers have joined Meta’s new “AI Superintelligence” team, including:
Trapit Bansal: A key figure in OpenAI’s reinforcement learning and co-creator of the o1 reasoning model.
Lucas Beyer, Alexander Kolesnikov, Xiaohua Zhai: Zurich-based researchers, formerly of Google, specializing in computer vision and multimodal AI.
Shengjia Zhao, Jiahui Yu, Shuchao Bi, Hongyu Ren: Experts in post-training and reasoning models, per The Information.
Meta’s Strategy: CEO Mark Zuckerberg is personally recruiting top talent, offering multimillion-dollar packages with equity, long-term incentives, and high autonomy. Reports of $100 million signing bonuses, mentioned by Altman on the Uncapped podcast, were disputed by Meta as exaggerated, though lucrative offers are confirmed.
Additional Hires: Meta also recruited Scale AI CEO Alexandr Wang (after a $14.3 billion investment), Google DeepMind’s Jack Rae, and Sesame AI’s Johan Schalkwyk.
OpenAI’s Response:
Compensation Overhaul: Chen’s memo outlined “recalibrating compensation” and exploring “creative ways to recognize and reward top talent,” such as performance-based bonuses and equity adjustments.
Leadership Engagement: Chen, Altman, and other executives are working “around the clock” to retain staff with offers from Meta, emphasizing fairness in compensation adjustments.
Cultural Push: Chen urged employees to focus on OpenAI’s mission of achieving artificial general intelligence (AGI), calling Meta’s efforts a “side quest” compared to the “main quest” of groundbreaking AI research.
Context and Implications
AI Talent War: The competition for AI researchers is at a fever pitch, with companies like Meta, Google, and Anthropic vying for top talent. Meta’s $14.3 billion stake in Scale AI and failed bids for startups like Safe Superintelligence and Perplexity AI underscore its aggressive push to catch up in the AI race after Llama 4’s lukewarm reception in April 2025.
OpenAI’s Challenges: The loss of eight researchers, including foundational contributors to models like o1, threatens OpenAI’s lead in AI reasoning. Altman’s claim that “none of our best people” have left was contradicted by Bansal’s departure, a significant blow given his role in reinforcement learning.
Meta’s Ambition: Meta’s new Superintelligence team, led by Wang, aims to rival OpenAI’s GPT and Google’s DeepMind, focusing on enterprise tools, consumer products, and AI agents. The hiring spree follows Meta’s reported interest in voice cloning startup Play AI, signaling a broader AI strategy.
Industry Trends: The AI talent market is likened to professional sports, with top researchers commanding multimillion-dollar packages. Google DeepMind enforces 6–12-month noncompetes, paying sidelined researchers full salaries, while Anthropic boasts an 80% retention rate, per SignalFire’s 2025 State of Talent Report.
Critical Analysis
Skepticism on Bonuses: While Altman cited $100 million signing bonuses, Meta’s pushback and Lucas Beyer’s denial (“No, we did not get 100M sign-on”) suggest these figures may include equity and long-term incentives, not cash bonuses alone. Still, Meta’s offers are undeniably lucrative, with Zuckerberg reportedly offering $2 million/year floors.
OpenAI’s Vulnerability: OpenAI’s focus on frequent product launches, criticized by Chen as a distraction from AGI, may have demoralized staff, making Meta’s offers more appealing. The loss of Zurich-based researchers suggests Meta is targeting OpenAI’s global talent hubs.
Ethical Concerns: Chen’s emphasis on “fairness” in compensation adjustments may struggle to compete with Meta’s aggressive tactics, raising questions about OpenAI’s ability to retain talent without matching exorbitant offers. X posts, like @TheTranscript_, highlight the intensity of Meta’s strategy, while @mujifren notes the broader AI talent rush amid layoffs in other tech sectors.
Market Impact: Meta’s hiring could accelerate its AI capabilities, especially in reasoning models, but its retention rate (64%) lags behind Anthropic’s (80%), per SignalFire. OpenAI’s planned open-source reasoning model release in summer 2025 could further pressure Meta.
0 Comments