Zuckerberg offers US$100 million bonuses and buys into Scale AI as Meta builds Superintelligence Labs
Meta shares reached an all-time high of US$747.90 during midday trading on Monday, according to CNBC, surpassing the company’s previous peak in February.
The stock movement follows rising investor interest in Meta’s artificial intelligence expansion, despite an estimated US$1bn in annual expenses tied to its recruitment strategy.
Bank of America, as reported by CNBC, projected that hiring 50 top AI researchers could cost Meta US$1bn annually, with average compensation estimated at US$20m per researcher.
According to CNBC, the hiring effort has included signing bonuses of up to US$100m, as claimed by OpenAI CEO Sam Altman.
The investor response remains strong. Between the market close on Friday, June 6, and Monday, Meta shares gained nearly 6 percent compared to a 3.4 percent rise in the S&P 500.
Bloomberg News had first reported during that weekend that Meta was in talks to invest in Scale AI — a move that later materialized as a US$14.3bn deal for a 49 percent stake.
Meta’s new AI division, Superintelligence Labs (MSL), is led by former Scale AI CEO Alexandr Wang and ex-GitLab CEO Nat Friedman.
According to CNBC, the division brings together Meta’s foundation models, product teams, and Fundamental AI Research (FAIR), as well as a new lab focused on developing the next generation of models.
In an internal memo obtained by CNBC, Zuckerberg wrote that “developing superintelligence is coming into sight” as AI progress accelerates. He described it as “the beginning of a new era for humanity” and said he was fully committed to ensuring Meta leads the way.
Meta has also attempted to acquire Safe Superintelligence, but its founder, Ilya Sutskever, declined the offer. However, Meta hired the company’s CEO Nat Friedman and partner Daniel Gross.
The broader recruitment strategy has targeted researchers from OpenAI, Google, and Anthropic.
According to Reuters, recent hires from OpenAI include Shengjia Zhao, Jiahui Yu, Shuchao Bi, and Hongyu Ren.
Earlier in the same week, Meta also brought on Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai from OpenAI’s Zurich office, as reported by the Wall Street Journal.
In the same memo, Zuckerberg highlighted the resumes of new AI hires:
-
Trapit Bansal, co-creator of o-series models at OpenAI
-
Shuchao Bi, co-creator of GPT-4o voice mode
-
Huiwen Chang, known for MaskGIT and Muse at Google Research
-
Ji Lin, contributor to GPT-4 and Llama models
-
Joel Pobar, former Meta engineer now returning from Anthropic
-
Pei Sun, key figure behind Waymo’s perception models
-
Johan Schalkwyk, ex-Google Fellow and early contributor to Sesame
-
Jiahui Yu and Shengjia Zhao, both with multimodal and synthetic data experience from OpenAI
Zuckerberg added that he was optimistic about the company’s direction, saying the new influx of talent and parallel model development would “set us up to deliver on the promise of personal superintelligence for everyone.”
Despite the escalating costs, Meta’s AI initiatives have already enhanced ad targeting and user engagement within its Family of Apps, according to CNBC.
However, its Llama large language model — currently in its fourth version — has yet to outperform competitors, with CNBC citing the LMArena.ai leaderboard as showing it trailing Google and OpenAI offerings.
Improving Llama is a key priority for the Superintelligence Labs unit.
Meta’s roadmap may also include premium AI features, business tools, and AI-enhanced smart glasses via its Reality Labs division.
Andrew Bosworth, Meta’s technology chief, told CNBC’s Closing Bell Overtime on June 20 that the AI talent market is “really incredible and kind of unprecedented in my 20-year career.”
