Anthropic Faces Strategic Pullback as Tech Giants Lean In-House; Reflection Unveils Beam, Workday Grapples with AI Liability

Major cloud companies Meta and Microsoft have sharply reduced their internal reliance on Anthropic's Claude AI, marking a significant strategic shift toward proprietary systems even as smaller players like Reflection push ambitious open-weight alternatives. Microsoft cut its planned annual spending on Anthropic from approximately $1 billion to around $667 million—a reduction of roughly one-third—according to reporting by The Information, with the company now steering employees toward GitHub Copilot for coding tasks, though engineers retain the option to use competing models. Meta undertook an even more aggressive pullback, reducing Claude Code users from roughly 60,000 employees to 30,000, pushing the workforce toward its internal Muse Code and MetaCode tools. Neither company has canceled contracts with Anthropic or restricted Claude's availability to external customers, but the reductions signal a tactical realignment in how the world's largest technology firms allocate AI resources internally.