Cloudflare, a leading US-based internet infrastructure and cybersecurity company, has announced a significant reduction in its global workforce, affecting over 1,100 employees. This move represents approximately 20% of the company's total workforce, which stood at 5,156 full-time employees at the end of 2025.
CEO Matthew Prince addressed the decision during an earnings call, stating, "This wasn't an easy decision, but it's the right decision." He explained that the company is transitioning to an "agentic AI-first operating model," a strategic shift that has fundamentally altered how Cloudflare operates.
AI Transformation Drives Restructuring
According to Prince, the dramatic increase in AI utilization, which has grown by 600% in the last three months, enables smaller teams to manage tasks that previously required considerably larger workforces. This transformation necessitates a restructuring of roles, meaning some existing positions no longer align with Cloudflare's future priorities.
The company clarified in a blog post that the layoffs are a result of organizational restructuring and process changes, not due to employee underperformance or temporary budget constraints. This distinction is crucial, especially given Cloudflare's strong financial performance.
Despite Strong Earnings, Strategic Shift Persists
In the first quarter, Cloudflare surpassed Wall Street expectations, reporting revenue of $639.8 million against forecasts of $621.9 million. Its adjusted profit also exceeded projections, reaching 25 cents per share. Despite these robust earnings, the company is proceeding with the workforce reduction as part of its long-term strategic evolution towards an AI-centric operational framework.
Cloudflare anticipates incurring between $140 million and $150 million in severance pay during the second quarter as a result of these layoffs. The company's decision highlights a broader trend in the tech industry where advancements in artificial intelligence are prompting significant changes in workforce structures and operational models.