The admission centered on two core failures:

• Disruptive Restructuring Execution: Zuckerberg described the broad AI reorganization as “disruptive” for many employees, a tacit acknowledgment that the speed and forced nature of the changes had taken a significant human toll .
• The Llama 4 Credibility Crisis: A more glaring mistake involved Meta’s flagship open-source model. Shortly after launch, independent testers discovered the publicly released version of Llama 4 significantly underperformed the custom-tuned variant Meta submitted to benchmark leaderboards. LMArena, a prominent AI benchmark, confirmed the discrepancy and updated its policies as a result. The incident severely damaged Meta’s reputation as a champion of open-source AI . Yann LeCun, who stepped down as Meta’s chief AI scientist, later told the Financial Times the company had “fudged” some Llama 4 performance metrics .

The Scale of Human Disruption: Layoffs and Reassignments

Behind the memo’s diplomatic language lies one of the most sweeping corporate restructurings in the tech industry’s recent history. In May 2026, Meta executed a multi-phase plan that touched roughly 20% of its workforce .

• Layoffs: Meta cut approximately 8,000 positions, representing 10% of its global workforce. The cuts, which began on May 20 and were rolled out in three phases starting at 4 a.m. local time for affected employees, hit engineering and product teams particularly hard. Integrity, cybersecurity, and content design teams were also significantly affected .
• Forced Reassignments: Days before the layoffs, Meta’s chief people officer Janelle Gale informed 7,000 employees they were being moved into new AI-focused divisions. These reassignments were explicitly non-optional. The new structures included teams working on an internal AI agent called Hatch and cloud infrastructure, with a mandate for “AI-native design structures” and a flattened management hierarchy .
• Closed Roles: Compounding the impact, around 6,000 open roles were quietly closed, eliminating internal mobility options for many staff. When combined, the layoffs, reassignments, and closed roles meant approximately 15,000–20,000 people were either leaving or being pushed into new, unfamiliar jobs .

Employee Morale and Internal Friction

The upheaval has bred a deep morale crisis inside the company, even as Meta continues to post strong quarterly profits. Employees have voiced concerns over several factors :

• AI-driven restructuring and the forced nature of role changes, with some internally describing the process as being “drafted” into new positions .
• Implementation of workplace monitoring software.
• Changes to compensation structures.
• A pervasive sense of anxiety and declining trust in leadership, with many questioning job security despite Zuckerberg’s promise of no further company-wide layoffs in 2026 .

Friction wasn’t limited to the rank-and-file. The arrival of 28-year-old Alexandr Wang as Chief AI Officer led to reported clashes with longtime Meta veterans Chris Cox and Andrew Bosworth. Meta has officially denied these conflicts, but the reported tension highlighted the cultural collision between established social-media leadership and the aggressive, start-up-style AI teams brought in to accelerate progress .

Record AI Spending Amid Wall Street Doubts

At the financial heart of this turmoil is a spending plan of unprecedented scale. Meta raised its 2026 capital expenditure guidance twice, first to a range of $115–$135 billion in January, and then to $125–$145 billion during its Q1 earnings call in April .

This figure represents a roughly 73% jump over the $72.2 billion spent in 2025, and nearly double the company’s entire 2025 capex. The spending is driven by massive data center expansions, rising component prices, and infrastructure commitments, including a multibillion-dollar partnership with Amazon Web Services .

The market’s reaction was swift and negative. On April 29, 2026, following the capex revision, Meta’s stock slid by 6–7% in after-hours trading. Analysts and investors expressed deepening “concern about [the] AI spending spree,” drawing harsh comparisons to the billions lost in Meta’s metaverse pivot . So great is the financial appetite that the Financial Times reported Meta’s management is considering a rights offering—a paid-in capital increase—to help fund the AI buildout .

The Competitive Pressure Driving the Overhaul

These drastic measures are a direct response to a bruising competitive reality. Meta’s restructuring is explicitly designed to close the gap with frontier AI leaders OpenAI, Alphabet (Google/DeepMind), and Anthropic .

Meta’s position weakened after it poured billions into a next-generation foundation model, codenamed Avocado and officially called Muse Spark, only to delay its release due to performance issues. The company found the model did not master internal tests on reasoning, coding, and writing sufficiently .

Combined with the Llama 4 benchmark controversy, these setbacks widened the gap with rivals deploying sophisticated closed-source models like OpenAI’s GPT series and Anthropic’s Claude. For a company that has staked its AI brand on open-source leadership, the credibility hit was especially damaging . The entire reorganization—from the $145 billion in infrastructure to the forced human overhaul—is a bet that Meta can regain its footing in a race where falling further behind is not an option.

The admission centered on two core failures:

• Disruptive Restructuring Execution: Zuckerberg described the broad AI reorganization as “disruptive” for many employees, a tacit acknowledgment that the speed and forced nature of the changes had taken a significant human toll .
• The Llama 4 Credibility Crisis: A more glaring mistake involved Meta’s flagship open-source model. Shortly after launch, independent testers discovered the publicly released version of Llama 4 significantly underperformed the custom-tuned variant Meta submitted to benchmark leaderboards. LMArena, a prominent AI benchmark, confirmed the discrepancy and updated its policies as a result. The incident severely damaged Meta’s reputation as a champion of open-source AI . Yann LeCun, who stepped down as Meta’s chief AI scientist, later told the Financial Times the company had “fudged” some Llama 4 performance metrics .

The Scale of Human Disruption: Layoffs and Reassignments

Behind the memo’s diplomatic language lies one of the most sweeping corporate restructurings in the tech industry’s recent history. In May 2026, Meta executed a multi-phase plan that touched roughly 20% of its workforce .

• Layoffs: Meta cut approximately 8,000 positions, representing 10% of its global workforce. The cuts, which began on May 20 and were rolled out in three phases starting at 4 a.m. local time for affected employees, hit engineering and product teams particularly hard. Integrity, cybersecurity, and content design teams were also significantly affected .
• Forced Reassignments: Days before the layoffs, Meta’s chief people officer Janelle Gale informed 7,000 employees they were being moved into new AI-focused divisions. These reassignments were explicitly non-optional. The new structures included teams working on an internal AI agent called Hatch and cloud infrastructure, with a mandate for “AI-native design structures” and a flattened management hierarchy .
• Closed Roles: Compounding the impact, around 6,000 open roles were quietly closed, eliminating internal mobility options for many staff. When combined, the layoffs, reassignments, and closed roles meant approximately 15,000–20,000 people were either leaving or being pushed into new, unfamiliar jobs .

Employee Morale and Internal Friction

The upheaval has bred a deep morale crisis inside the company, even as Meta continues to post strong quarterly profits. Employees have voiced concerns over several factors :

• AI-driven restructuring and the forced nature of role changes, with some internally describing the process as being “drafted” into new positions .
• Implementation of workplace monitoring software.
• Changes to compensation structures.
• A pervasive sense of anxiety and declining trust in leadership, with many questioning job security despite Zuckerberg’s promise of no further company-wide layoffs in 2026 .

Friction wasn’t limited to the rank-and-file. The arrival of 28-year-old Alexandr Wang as Chief AI Officer led to reported clashes with longtime Meta veterans Chris Cox and Andrew Bosworth. Meta has officially denied these conflicts, but the reported tension highlighted the cultural collision between established social-media leadership and the aggressive, start-up-style AI teams brought in to accelerate progress .

Record AI Spending Amid Wall Street Doubts

At the financial heart of this turmoil is a spending plan of unprecedented scale. Meta raised its 2026 capital expenditure guidance twice, first to a range of $115–$135 billion in January, and then to $125–$145 billion during its Q1 earnings call in April .

This figure represents a roughly 73% jump over the $72.2 billion spent in 2025, and nearly double the company’s entire 2025 capex. The spending is driven by massive data center expansions, rising component prices, and infrastructure commitments, including a multibillion-dollar partnership with Amazon Web Services .

The market’s reaction was swift and negative. On April 29, 2026, following the capex revision, Meta’s stock slid by 6–7% in after-hours trading. Analysts and investors expressed deepening “concern about [the] AI spending spree,” drawing harsh comparisons to the billions lost in Meta’s metaverse pivot . So great is the financial appetite that the Financial Times reported Meta’s management is considering a rights offering—a paid-in capital increase—to help fund the AI buildout .

The Competitive Pressure Driving the Overhaul

These drastic measures are a direct response to a bruising competitive reality. Meta’s restructuring is explicitly designed to close the gap with frontier AI leaders OpenAI, Alphabet (Google/DeepMind), and Anthropic .

Meta’s position weakened after it poured billions into a next-generation foundation model, codenamed Avocado and officially called Muse Spark, only to delay its release due to performance issues. The company found the model did not master internal tests on reasoning, coding, and writing sufficiently .

Combined with the Llama 4 benchmark controversy, these setbacks widened the gap with rivals deploying sophisticated closed-source models like OpenAI’s GPT series and Anthropic’s Claude. For a company that has staked its AI brand on open-source leadership, the credibility hit was especially damaging . The entire reorganization—from the $145 billion in infrastructure to the forced human overhaul—is a bet that Meta can regain its footing in a race where falling further behind is not an option.