Google caps Meta's Gemini AI access as capacity crunch hits 2 rivals

Google told Meta around March it could not fulfill the full Gemini AI capacity the social media company had sought to purchase, the Financial Times reported, disrupting several of Meta's internal AI projects.

"Even the largest tech companies are struggling to secure enough computing power to support surging demand for advanced models and AI services," the FT reported, citing people familiar with the matter.

Google Cloud generated $20 billion in revenue in the first quarter ended March, but Chief Executive Officer Sundar Pichai said computing power constraints prevented even higher growth and contributed to the cloud unit's backlog nearly doubling quarter on quarter. Meta has since encouraged staff to be more efficient with AI tokens, the units that measure AI usage, as a result of the restrictions.

The move marks an escalation in Big Tech AI rivalry, with Google leveraging its dominant position in cloud-based AI infrastructure to limit a competitor's access. For Meta, losing full access to a leading AI model could slow its product roadmap, while for Google, the capacity crunch highlights the immense capital demands of the AI arms race.

The Infrastructure Bottleneck

Companies across the technology sector have spent tens of billions of dollars on chips, data centers, and power, yet supply still cannot keep pace with demand. Google's decision to ration Gemini access — affecting Meta more acutely than other clients due to the scale of its requests — offers a rare window into the infrastructure pressures building across the AI industry.

Nvidia's H100 and B200 graphics processing units remain the most sought-after chips for AI training and inference, with lead times stretching months for large orders. Microsoft, Amazon, and Google have each committed more than $50 billion in combined data center capital expenditure for 2026, according to company filings, yet cloud capacity remains constrained across all three hyperscalers.

What It Means for Investors

For Alphabet, the capacity constraints present a dual-edged narrative. Demand for Google Cloud's AI services is so strong that the company cannot build infrastructure fast enough — a problem that supports the bull case for GOOGL shares. But rationing a key customer like Meta risks pushing one of the world's largest AI spenders toward building more of its own infrastructure or deepening ties with rival cloud providers such as Microsoft Azure or Amazon Web Services.

Meta has been investing heavily in its own AI research, including its Llama family of large language models. Losing full access to Gemini could accelerate those internal efforts, though building competitive frontier models requires tens of thousands of GPUs and months of training time — resources that even Meta, with its more than $80 billion in annual capital expenditure, cannot deploy overnight.

Alphabet shares trade at about 22 times forward earnings, while Meta trades at about 18 times. The divergence partly reflects the market's view that Google's cloud business has more room to grow, but the capacity constraints introduce a new variable: if Google cannot serve demand, that growth may slow.

This article is for informational purposes only and does not constitute investment advice.