Meta 正在进入新一轮裁员，尽管其业务依然高度获利，财务与士气之间形成了鲜明失衡。公司计划在 5 月 20 日星期三裁减约 10% 的员工，约 8,000 人，这将增加过去 4 年宣布的约 25,000 人裁员。WIRED 采访了十多名前任与现任员工，他们描述了一种充满恐惧、愤怒与疲惫的氛围，这不仅源于裁员，还来自薪酬差距扩大、官司失利、被迫调整职务，以及用于 AI 训练、追踪员工活动的新软体。

主要抱怨集中在薪酬、监控和重组。2 月，Meta 在前一年削减 10% 之后，又把年度加薪中的股票部分再削减 5%，而去年员工薪酬总中位数则从 2024 年的 $417,400 降至 $388,200。Meta 的股价今年下跌约 5%，尽管公司在今年前 3 个月获利近 $27 billion，并将第一季度支出提高 35% 至 $33.4 billion。据报导，Zuckerberg 还曾向一些顶尖 AI 研究人员开出每年高达 $100 million 的报酬，同时把 2026 年资本支出指引上调 $10 billion，至 $125 billion-$145 billion。与此同时，至少 1,000 名顶尖工程师被调入新的 Applied AI Engineering 部门，而美国员工则被告知必须接受名为 Model Capability Initiative 的监控软体，以收集打字与点击资料供 AI 模型使用。

其后果不只影响士气，也延伸到劳工组织、隐私反弹与战略风险。英国员工正试图组建工会，美国员工已发起抗议与请愿，反对这套追踪软体，而有人表示公司最近的举措正在损害招募。加州与新墨西哥州的州法院裁决也加剧了不安，Meta 被认定需对造成近 $380 million 损害与民事罚款的伤害负责。Zuckerberg 则主张 AI 会加快工作速度，并表示过去需要几个月、数十名员工才能完成的任务，现在可由 1 或 2 人在 1 周内完成，但许多员工认为这是在警告自动化将取代工作。部分团队，尤其是 frontier-AI 团队，依然感到振奋；然而更广泛的氛围是不确定，员工正忙著证明自己应该在裁员落下前被保留下来。

Meta is entering a new round of layoffs even as its business remains highly profitable, creating a stark mismatch between finances and morale. The company plans to cut about 10 percent of staff, or nearly 8,000 people, next Wednesday, May 20, adding to roughly 25,000 cuts announced over the past 4 years. WIRED spoke with more than a dozen current and former employees who described an atmosphere of fear, anger, and exhaustion, driven not only by job cuts but also by widening pay gaps, court losses, forced role changes, and new software that tracks employee activity for AI training.

The main complaints center on compensation, surveillance, and restructuring. In February, Meta cut the stock portion of annual raises by another 5 percent after a 10 percent reduction the year before, while median total compensation fell to $388,200 last year from $417,400 in 2024. Meta’s shares are down about 5 percent this year, even as the company posted nearly $27 billion in profit in the first 3 months of this year and raised first-quarter expenses 35 percent to $33.4 billion. Zuckerberg also increased 2026 capital expenditure guidance by $10 billion to $125 billion-$145 billion, after reportedly offering some top AI researchers as much as $100 million a year. Meanwhile, at least 1,000 top engineers were moved into a new Applied AI Engineering division, and U.S. employees were told they must accept monitoring software, called Model Capability Initiative, to capture typing and clicking data for AI models.

The consequences reach beyond morale and into labor organizing, privacy backlash, and strategic risk. Employees in the UK are trying to unionize, U.S. workers have launched protests and petitions against the tracking software, and some say the company’s latest actions are hurting recruiting. State court verdicts in California and New Mexico also added to unease, with Meta found liable for harms that led to nearly $380 million in damages and civil penalties. Zuckerberg has argued that AI will speed up work, saying tasks that once took months and dozens of workers can now take 1 or 2 people and be done in 1 week, but many employees see that as a warning that automation will replace jobs. Some teams, especially frontier-AI groups, remain excited, yet the broader mood is one of uncertainty, with workers racing to prove they should be spared before the cuts land.