Meta's AI push under strain as Wang voices concern and leadership churn continues

Alexandr Wang, the 28-year-old founder of Scale AI who joined Meta earlier this year after Meta paid more than $14 billion for a 49% stake in his startup, is described by Financial Times sources as telling associates that Mark Zuckerberg’s micromanagement of Meta’s AI push is "suffocating." The remarks, reported amid broader upheaval inside the company, underscore the strain behind Meta’s high-profile bet on artificial intelligence and the fragility of its execution at scale.

Meta has positioned Wang as the public face of its AI reset and has placed him at the helm of the company’s secretive TBD Lab, which is charged with building a flagship model codenamed Avocado. The move followed a sweeping hiring spree and expensive incentives intended to accelerate the company’s AI ambitions, including the acquisition deal for Scale AI. While Wang’s appointment signaled confidence in the company’s path forward, multiple people familiar with the matter said privately that his effectiveness in managing a sprawling research and product organization at Meta’s scale is being questioned by some colleagues.

![Alexandr Wang]https://nypost.com/wp-content/uploads/sites/2/2025/12/alexandr-wang-co-founder-chief-117497798.jpg?quality=75&strip=all&w=1024

Inside Meta, the tensions have grown alongside a string of setbacks that analysts and employees say reflect a rapid, sometimes disjointed push to outpace rivals in AI. The botched rollout of Llama 4, Meta’s eagerly anticipated AI model, became a punching bag for critics who argued the company lagged on coding and complex reasoning benchmarks and even accused Meta of attempting to game leaderboards by submitting a customized variant for rankings. The episode was cited by FT as a symbolic blow to Zuckerberg’s pledge to transform Meta into an AI powerhouse on a compressed timetable.

Cited in interviews and internal briefings, Wang’s role as the chief architect of Meta’s AI strategy has come under scrutiny as the company confronted the broader challenge of aligning disparate product teams, data pipelines, and testing regimes. Some Meta insiders said the organization struggled with fragmentation, with teams pursuing their own product roadmaps rather than a cohesive, company-wide strategy that could reliably scale AI across surfaces.

![Mark Zuckerberg]https://nypost.com/wp-content/uploads/sites/2/2025/12/meta-ceo-mark-zuckerberg-speaks-114909830_f9d497.jpg?w=1024

Leadership strain extended beyond product development. Nat Friedman, the former GitHub chief brought in to integrate AI models into Meta’s products, faced mounting pressure to deliver quickly, frustrating members of his team who felt products were rushed to beat competitors. The FT described these dynamics as part of a broader pattern of executive churn that has unfolded alongside a wave of departures and reorganizations at the company.

The turnover has been stark. Jennifer Newstead, Meta’s longtime chief legal officer, left for Apple; John Hegeman, the chief revenue officer, announced plans to depart for a startup. Yann LeCun, the acclaimed AI scientist and Turing Award winner, said he would leave to pursue a new AI initiative after reportedly objecting to reporting to Wang and seeing his research priorities trimmed. Other seasoned hires in business AI roles did not endure long either, including Clara Shih, recruited from Salesforce to lead business AI, who left within a year. Amid these exits, Meta carried out large-scale layoffs of AI staff, with roughly 600 workers cut in an effort to speed decision-making and focus resources more sharply.

With those changes, Meta’s spending on AI remains extraordinarily aggressive. The company has signaled that AI capital expenditures could approach $70 billion this year, with projections indicating costs could top $100 billion annually as the push accelerates. The outsized investment has drawn scrutiny from investors, who watched Meta’s stock slip amid fears about free cash flow in a technology environment characterized by high spending on infrastructure and research.

Meta has pushed back against some characterizations of internal upheaval. The company said it routinely experiments with different AI model variants and that public leaderboards can be misleading or "easily gameable." It also noted that earlier chatbot policy concerns cited in coverage were "erroneous and inconsistent" with its rules. Still, interviews with current and former employees painted a picture of a company racing to scale a frontier technology while contending with data quality issues, coordination gaps among teams, and the complexity of shepherding frontier AI research into consumer products.

As the AI agenda remains central to Meta’s strategy, observers say the company will need to reconcile its ambitious timetable with the realities of building and integrating advanced models at scale. The question for investors and workers alike is whether the current leadership shakeup and the insistence on rapid iteration will yield durable, cohesive product improvements or ongoing disruptions that undermine morale and execution. In the near term, Meta’s path will continue to be defined by how tightly it can align its top researchers, engineers, and product teams around a shared vision for AI-driven experiences across its platforms.

![AI concept]https://nypost.com/wp-content/uploads/sites/2/2025/12/large-language-model-meta-ai-21740233.jpg?w=1024