Meta Reportedly Projected Spending $10 Billion on Anthropic AI Services Before Shifting Strategy

Meta reportedly projected spending up to $10 billion on Anthropic's AI services around April this year before shifting its focus to internal models to reduce costs. This shift led Meta to end its internal "tokenmaxxing" competition and update its AI spending guidance in June. This strategic pivot highlights the massive costs associated with relying on third-party AI APIs and underscores the industry trend of tech giants prioritizing proprietary models. Additionally, Meta's reduced reliance on Anthropic could negatively impact Anthropic's annualized revenue and its future IPO prospects. Meta initially tested Anthropic's AI for its upcoming "Hatch" consumer agent platform, but the official release will be powered by Meta's own internal models. The transition away from external APIs helped Meta adjust its full-year AI application spending guidance to a lower "billions of dollars" range in June.