Google has placed usage restrictions on Meta Platforms' access to its Gemini artificial intelligence models, as surging demand outpaces the available computing capacity even for the world's largest tech companies. The news, first reported by the Financial Times and later confirmed by Tech in Asia, sent Google stock modestly higher and highlighted the intense strain on AI infrastructure.
According to reports, Meta requested significantly more Gemini computing capacity than Google could provide. The shortfall disrupted some of Meta's internal projects, including content moderation and scam detection, where Google's AI had reportedly outperformed Meta's own systems. In response, Meta has instructed employees to use AI resources more efficiently and is shifting more workloads to its proprietary Muse Spark model, aiming to reduce reliance on outside providers.
Wedbush Securities analyst Matt Bryson warned that the situation is the latest evidence that demand for computing power continues to exceed supply, even after billions of dollars in infrastructure investments. He also raised concerns about the risks of relying on competitors for resource allocation, a challenge that could affect other AI model builders like Anthropic that depend on Google's cloud services.
Alphabet CEO Sundar Pichai previously acknowledged near-term capacity constraints and noted that Google Cloud's backlog has expanded substantially. The tension between Meta's record $135 billion AI spending plan for 2026 and Google's inability to fulfill its requests underscores how enterprise AI adoption is creating unprecedented demand for chips, data centers, and specialized hardware.
