Media reports reveal that Meta founder Mark Zuckerberg published an article this month criticizing AI labs like Anthropic for attempting to monopolize power, yet Meta internally once projected annual spending on Anthropic AI models could reach $10 billion, and currently still spends hundreds of millions of dollars monthly on Anthropic tools, making it one of Anthropic's largest paying customers. This phenomenon reflects the complex "frenemy" landscape in the Silicon Valley AI race. Meta engineers extensively use Anthropic's Claude Code coding tool, and its upcoming AI product, Hatch, also utilized Anthropic models during its development and testing phases. Simultaneously, Meta is striving to develop its own AI model, "Watermelon," and launch its proprietary coding tool, Muse Code, to reduce reliance on external providers.