Meta's top-paid AI hire unhappy with Mark Zuckerberg's leadership: Report

Meta’s aggressive push to emerge as a global leader in artificial intelligence is facing internal strain, with a new report suggesting growing unease among senior leaders over how the effort is being run.

According to a Financial Times report, Alexandr Wang — the billionaire founder of Scale AI who joined Meta earlier this year — has privately expressed frustration over CEO Mark Zuckerberg’s highly hands-on management style, describing it to associates as restrictive. Wang became a central figure in Meta’s AI strategy after the company invested over $14 billion for a 49% stake in his data-labeling startup.

Publicly positioned as a key architect of Meta’s AI reset, Wang is said to be struggling with the pace and structure of decision-making, as Zuckerberg retains close control over product direction and research priorities. People familiar with the situation told the FT that this approach has contributed to internal tension rather than speeding execution.

The reported friction comes amid broader upheaval inside Meta. Over the past year, the company has seen repeated layoffs, senior exits and pressure-filled AI launches, even as it ramps up spending at an unprecedented scale.

One internal setback highlighted in the report was the rollout of Meta’s Llama 4 large language model. Despite high expectations, the model reportedly lagged competitors in benchmarks related to coding and reasoning. Meta also faced criticism over allegedly submitting a customised version for public rankings. Employees cited gaps in training data, limited testing time and organisational silos as contributing factors.

Rather than slowing down, Zuckerberg has reportedly intensified efforts — greenlighting massive investments in AI infrastructure and launching a hiring spree across Silicon Valley, with compensation packages running into tens of millions of dollars.

Wang now leads a confidential internal unit tasked with developing Meta’s next-generation AI system, codenamed “Avocado”. However, the FT report notes that some employees have questioned whether his background in AI data services fully equips him to manage large-scale frontier research teams at Meta’s size.

Tensions have also been reported involving Nat Friedman, the former GitHub CEO overseeing AI product integration. Teams have reportedly faced pressure to ship features rapidly to keep pace with rivals such as OpenAI, sometimes at the cost of internal alignment and testing.

At the leadership level, Meta has seen a steady stream of departures. Chief legal officer Jennifer Newstead recently joined Apple, while chief revenue officer John Hegeman has announced plans to leave. Veteran AI scientist Yann LeCun is also expected to step away to pursue an independent research initiative, amid disagreements over reporting structures and research focus.

Despite the churn, Meta has continued to cut roles within its AI teams, framing the layoffs as a move to streamline operations.

All of this is unfolding against the backdrop of enormous spending. Meta’s AI capital expenditure is projected to reach at least $70 billion this year, with Zuckerberg signalling that annual costs could climb even higher — a trajectory that has unsettled investors concerned about long-term cash flow.

Meta has disputed aspects of the Financial Times report, saying it routinely tests multiple AI model variants and cautioning that public benchmarks can be misleading. The company also pushed back on suggestions of widespread internal dysfunction.

As Meta races to keep up in the global AI arms race, the report suggests that questions around leadership style, speed and organisational clarity may be as critical as the technology itself.