Meta CEO Mark Zuckerberg told employees on Friday that the company had made “mistakes” in its rapid pivot to artificial intelligence, speaking internally one day after completing its largest round of job cuts since 2022. The layoffs eliminated roughly 8,000 positions, approximately 10 percent of Meta’s total workforce, with affected workers in Singapore, the United Kingdom, and the United States notified in the early hours of their local morning.

Zuckerberg did not specify which decisions he regretted, but said the pace of AI-related change had made it difficult to predict which teams and skill sets the company would need. He added that further company-wide cuts were not planned for 2026.

Beyond the reductions, Meta announced that approximately 7,000 employees are being reassigned into newly created AI-focused divisions, including teams dedicated to applied AI engineering, an Agent Transformation Accelerator unit, and a central analytics group.

Meta projected capital expenditures of between $125 billion and $145 billion for 2026, more than double the previous year’s spending. The investment is directed toward AI data centres, custom silicon development, and training infrastructure for its next-generation Llama 4 models.

The scale of the cuts drew sharp reactions from technology workers and labour advocates across Singapore, the UK, and the US. Singapore’s government said it expected Meta to fulfil all local employment obligations. Trade unions in Ireland said they were seeking urgent meetings with management.

Zuckerberg pledged to increase investment in team-building, raising budgets for off-site events and planning a company-wide hackathon in July focused on collaborative AI model development. Meta’s shares rose slightly on Friday. The shifts compare with Microsoft’s own AI model investments at Build 2026. Meta investor relations published supporting financial data. The Nvidia RTX Spark chip powering Meta’s hardware push and Anthropic’s IPO both shaped the AI arms race context.