August 20, 2025 – Amid a heated artificial intelligence (AI) race, Meta (formerly Facebook) is once again in the spotlight. After months of aggressively recruiting top global AI talent, Meta this week announced a major reorganization of its AI division, splitting the newly formed Meta Superintelligence Labs into four independent teams. The move is being described as the company’s fourth AI strategic adjustment in six months. Industry analysts say it both reflects Meta’s ambitious AI agenda and exposes internal pressure and resource‑allocation challenges brought by rapid expansion. As the AI talent war intensifies, these moves could reshape Silicon Valley’s competitive landscape and prompt deeper debates over AI ethics, employment, and global technological balance.
Meta’s AI push began after the 2023 breakout of ChatGPT, with founder Mark Zuckerberg taking a personal lead to make AI a core company focus. In the first half of 2025, Meta’s investment in AI infrastructure reportedly reached between US$64 billion and US$72 billion, far exceeding the prior year. Part of that spending went to building data centers and 1‑gigawatt‑class superclusters, in some cases using rapid, tent‑style construction to accelerate model training. At the same time, Meta launched an unprecedented hiring spree targeting core talent from competitors such as OpenAI, Google DeepMind, Anthropic and Apple. Compensation packages have reportedly reached eye‑watering levels: signing bonuses up to US$100 million, nine‑figure salaries, and equity incentives that can push total value beyond US$200 million. For example, Meta’s package to recruit Ruoming Pang, the former head of Apple’s large language model team, was reportedly worth hundreds of millions of dollars in multi‑year equity and bonuses.
The recruitment drive peaked in 2025. Reports say Meta hired at least seven to eight top researchers from OpenAI, including Lucas Beyer, Alexander Kolesnikov, Xiaohua Zhai, Yuanzhi Li, Trapit Bansal, Shengjia Zhao and Huiwen Chang, all noted for work in visual AI and model development. Meta also recruited Ruoming Pang from Apple (his former Apple team numbered around 100). From Scale AI and GitHub and other firms, Meta lured senior figures such as former Scale AI CEO Alexandr Wang, who joined as Chief AI Officer; former GitHub CEO Nat Friedman to lead AI products; and Safe Superintelligence founder Daniel Gross into product roles. Other notable hires include former Google Fellow Johan Schalkwyk and DeepMind’s Pei Sun. These additions rapidly expanded Superintelligence Labs, which is focused on developing general artificial intelligence (AGI) and pursuing "personal superintelligence" realized through devices like AR glasses.
Discussions on X (formerly Twitter) amplified the so‑called "AI talent war." One post claimed Meta’s hiring budget reached US$72 billion and that poaching from OpenAI and Anthropic was particularly aggressive, including offers of US$25 million to recruit 24‑year‑old AI dropout Matt Deitke. Another post alleged Meta had recruited nearly half the members of an elite Chinese lab, stirring debate about national security and global talent flows. Elon Musk’s xAI also entered the conversation, claiming it could attract engineers without matching such high pay, but Meta has countered with concrete offers under a stated "all‑in U.S." strategy. At a July 23, 2025 AI summit, former President Donald Trump urged tech giants to stop recruiting from countries like India and instead prioritize American workers—an intervention that indirectly affected global hiring plans. Anthropic CEO Dario Amodei reportedly declined to match a US$200 million offer in an internal memo, stressing culture and mission over money and writing, "If Mark Zuckerberg pins your name on a dartboard, that doesn't mean you should be paid ten times what the person next to you makes."
But behind the aggressive recruiting, strains have emerged inside Meta’s AI organization. On August 19, 2025, Meta internally announced that Superintelligence Labs would be split into four teams: one mysterious unit dubbed the "TBD Lab" focused on speculative frontier research; a product‑focused team including AR/VR integration; a team responsible for infrastructure optimization and data center expansion; and a fourth dedicated to AI safety and ethics to ensure models comply with global regulations. This marks the fourth reorganization in six months—the prior three occurred in February (research team consolidation), April (separation of product and infrastructure), and June (adding external talent to leadership)—each accompanied by departmental reshuffles and resource reallocation. Insiders say the latest reorganization stems from tensions created by rapid hiring: high‑paid newcomers clashing with existing team culture and uneven resource distribution leading to inefficiencies. Some former OpenAI hires have complained about Meta’s bureaucracy, while legacy employees have expressed resentment over perceived "newcomer privileges."
More worryingly, the reorganization may include headcount reductions. Reports say Meta is considering shrinking parts of its AI workforce overall; while no final decision has been announced, multiple employees have received internal notices hinting at role changes or "efficiency optimizations." This contrasts with Meta’s Q2 2025 results: research and development spending reached US$5.1 billion, up 21% year over year, primarily driven by AI hiring, even as net income growth slowed and the price‑to‑earnings ratio fell to 25.3, lagging Google’s 28.1. Analyst Sarah Kunst told CNBC, "Meta’s AI hiring spree may yield nothing — it’s another expensive moonshot for Silicon Valley." Indeed, investors have grown concerned: 2025 capital expenditures are projected at US$64–72 billion, nearly double prior levels, aimed at multi‑gigawatt superclusters and data centers the size of city blocks.
From an industry perspective, Meta’s moves have intensified an AI talent shortage. In the first half of 2025, startup funding for AI reached about US$73 billion, yet giants like Meta poaching with outsized pay have made it harder for smaller firms to compete for talent. OpenAI reportedly increased equity compensation to US$4.4 billion to retain staff, and CEO Sam Altman publicly criticized Meta’s "mercenary" hiring approach, saying "mission‑driven teams will beat mercenaries." Anthropic’s Dario Amodei told the Big Technology Podcast that many employees refuse to entertain conversations with Zuckerberg, prioritizing a mission to build safe, ethical AI. Google reportedly hired fewer than 24 employees from Meta, but talent mobility has become the new norm. Reddit and X threads show criticism from some former OpenAI staff—such as Joaquin Quiñonero Candela—who described Meta’s "fake deadline" recruiting tactics as "desperate."
The deeper motive behind the reorganization appears strategic. In an internal meeting, Zuckerberg framed Meta’s AI goal as "amplifying human potential," not replacing jobs. By splitting teams, the company aims to more efficiently iterate models like Llama 4 and to better respond to regulatory pressure, including the EU’s AI Act, which in 2025 banned certain high‑risk AI applications. Yet challenges remain: Meta’s metaverse initiatives have reportedly lost about US$60 billion, and if AI investments fail to deliver returns they could weigh on the stock. At the employee level, success of the reorganization will hinge on cultural integration: many new hires have Chinese or Chinese‑background ties, raising geopolitical concerns. An anonymous Meta engineer posted on Reddit, "High‑paid new hires arrive while the original team has to put in extra hours to adapt." If layoffs occur, they could trigger further departures, amplifying the talent war.
Looking ahead, Meta’s reorganization may signal a maturing phase for the AI industry: a shift from frantic expansion to more disciplined management. The company has pledged "tens of billions" for AI, but must demonstrate value, for instance by advancing toward AGI within Superintelligence Labs. Success could allow Meta to reduce reliance on advertising and dominate an AR/VR AI ecosystem; failure could render this another tech bubble. Either way, the upheaval is reshaping the global AI ecosystem and forcing the industry to reconsider sustainability and ethical boundaries.