Why Meta Lost Access to Google’s Gemini AI Models
According to a Financial Times report cited by Investing.com, Google restricted Meta Platforms—the parent company of Facebook—from accessing its Gemini artificial intelligence models after Meta requested more computing power than Google could provide. Around March 2026, Google reportedly informed Meta that it couldn’t meet the company’s full demand for Gemini resources. These restrictions remained in place as of late June 2026, slowing some of Meta's internal AI projects.
The Ripple Effect of Scarce Computing Power
Meta responded by urging employees to use AI resources more efficiently as part of a broader effort to control IT expenses. Other Google clients were reportedly affected by similar capacity limits, but Meta felt the greatest impact due to its exceptionally high demand.
This situation highlights the broader infrastructure challenges facing the AI sector. The demand for computing power continues to outpace the resources available, despite massive investments in semiconductors, data centers, and energy infrastructure.
How Google and Meta Are Responding
Google has worked to expand its own computing resources to keep pace. Earlier in June 2026, the company finalized an agreement to rent additional computing capacity from SpaceX, reportedly valued at about $920 million per month.
During Google’s first-quarter results in April, CEO Sundar Pichai stated that Google Cloud’s revenue had exceeded $20 billion for the first time. Backlog for signed but yet-to-be-delivered cloud contracts nearly doubled from the previous quarter, topping $460 billion. Pichai acknowledged that computing capacity remained limited in the short term, noting that cloud revenue could have been higher if Google had met all customer demand.
Meta’s Ambitious AI Investment—and a Shift in Strategy
Meta has continued to invest heavily in AI infrastructure, with CEO Mark Zuckerberg seeking to strengthen the company’s AI capabilities. Meta has committed to investing up to $600 billion in the United States by 2028 to expand its data center capacity. Internally, Meta used Google’s Gemini models for coding, customer service, ad tools, and content moderation. According to reports, the company recently began moving some workloads to its own Muse Spark model, decreasing its reliance on third-party AI models for certain tasks.
The bottom line: The competition for cloud and AI dominance is intensifying, and even the biggest tech names need to adapt. It turns out the digital world doesn’t run on magic—it's powered by data centers, and server space is becoming increasingly precious.
