Zuckerberg's $15 Billion Gamble: Meta's 'Superintelligence Labs' Set Off the Costliest AI Talent Battle in Silicon Valley

Mark Zuckerberg is renowned for taking bold risks that redefine industries, though some have resulted in failures. After investing $46 billion in the metaverse with limited returns, the Meta CEO is now making an even riskier investment in superintelligence AI.

This venture is particularly significant with fierce competition and transformative potential unlike anything Meta has pursued before.

With significant financial packages and infrastructure investments surpassing the metaverse initiative, Zuckerberg's foray into superintelligence AI marks Silicon Valley's most costly pursuit for talent, with the potential to reinforce Meta's stature as a leading tech firm or become a cautionary story of ambition clashing with reality.

The launch of Meta Superintelligence Labs signifies a strategic transformation for the social media giant. In an exclusive talk on The Information‘s TITV live-stream, Zuckerberg shared with founder Jessica Lessin that “the most exciting thing this year is observing initial signs of self-improvement in the models, indicating that developing super intelligence is now within reach.”

The objective compelled the company to overhaul its AI division, aiming for what Zuckerberg terms “personal super intelligence for everyone.” Forming the lab follows internal challenges within Meta’s AI division, including managerial shifts, employee turnover, and subpar product launches.

Instead of lightly tweaking existing systems, Zuckerberg chose a complete revamp, enlisting external leadership and rethinking the company’s AI development methodology.

At the heart of Meta’s superintelligence AI vision is a recruitment strategy causing ripples across the industry. Zuckerberg has launched a massive spending effort to set up the new lab, extending lucrative nine-figure offers to recruit top-tier researchers from companies like OpenAI, Google, Apple, and Anthropic.

When Lessin from The Information questioned reports of $100-$200 million compensation offers, Zuckerberg acknowledged the competitive climate, noting that “a lot of specifics reported aren't entirely accurate. However, the market is indeed competitive... there's a small group of exceptional researchers in demand by many labs.”

Meta’s recruitment strategy involves more than financial incentives. Zuckerberg commented that offering “the most computing resources per researcher gives a strategic edge, not only for conducting the work but also in attracting the best talent.” This reflects an understanding that in the superintelligence AI race, the concentration of talent holds more importance than sheer team numbers.

A major element of Meta’s talent strategy was acquiring leadership from Scale AI. In June, Meta invested $14.3 billion in the AI startup, acquiring a 49% stake, with Scale's CEO Wang and key employees joining Meta’s leadership.

At the age of 28, Alexandr Wang leads as Meta’s chief AI officer, heading the rebranded “Meta Superintelligence Labs.” Alongside a new research cohort and Scale AI deputies, the team signifies a reimagined approach to AI development.

Wang’s team integration highlights Meta’s fresh AI development approach, operating in an independent office space close to Zuckerberg, underlining the initiative’s priority.

The most significant development from Meta Superintelligence Labs is considering leaving behind an open-source model. Recently, top lab members and Wang brainstormed moving on from the open-source Behemoth model in favor of creating a closed model.

This consideration marks a new direction for Meta. Historically, Meta has released AI models publicly, advocating that open development accelerates AI advancement and is more accessible to developers.

The shift indicates concerns around competitive positioning in the AI space. Meta delayed releasing the Behemoth model due to suboptimal performance, prompting reevaluation of their strategy.

In addition to acquiring talent, Meta is investing in infrastructure to support its superintelligence goals. Zuckerberg unveiled that the company is “constructing multiple, multi-gigawatt data centers” using innovative building techniques, like “weatherproof tents” for faster deployment.

The scale of investment is vast. The Hyperion data center will grow to a five-gigawatt capacity over time, with a site footprint comparable to a substantial part of Manhattan.

These investments are backed by Meta’s solid financial standing, with Zuckerberg stating that “the funding will come from the company’s cash flow.”

Meta’s unique approach involves “personal superintelligence” over central AI systems. During the TITV interview, Zuckerberg explained that unlike other labs focusing on “automating productive work,” Meta prioritizes “aspects of life people value, like relationships, creativity, and enjoyment.”

This vision aligns with Meta’s hardware goals, especially its AR glasses project. In the same interview, Zuckerberg forecasted that “without AI glasses, individuals may face cognitive disadvantages,” describing scenarios where AI companions could “monitor life events and follow up accordingly.”

Meta's push toward superintelligence affects not only the company but also the broader industry. Meta’s recruitment tactics have instigated salary inflation, pressuring competitors to match or surpass Meta's offers to keep their talent.

Addressing competition at Sun Valley, Zuckerberg reflected on the competitive terrain, remarking, “we're not targeting anyone specifically. It’s my aim to familiarize myself with top industry researchers.”

This diplomatic stance belies a cutthroat competition for a limited pool of elite superintelligence AI talent. The pivot from open-source methods could herald a wider shift toward proprietary AI development across the industry.

Meta’s superintelligence project is redefining the company's future. Following the unfruitful metaverse endeavor, Zuckerberg's AI investments could well exceed $100 billion in coming years.

The fate of Meta Superintelligence Labs will likely shape not only Meta's future but also the AI industry's direction. With expectations of “talent withdrawal from those not chosen for Wang’s team,” the stakes are elevated.

Whether Meta's talent pursuits will lead to groundbreaking technologies envisioned by Zuckerberg remains uncertain. What is indisputable is that Silicon Valley's most costly talent rivalry has begun.