Google has reportedly restricted Meta’s access to its Gemini artificial intelligence models, citing a shortfall in computing capacity that the tech giant could not fulfill, according to a report by the Financial Times. This move underscores growing pressure on cloud infrastructure amid surging demand for generative AI.
The report states that Google informed Meta around March about its inability to provide the full amount of computing capacity requested for Gemini. Meta’s high demand reportedly made it one of the hardest hit customers, with the capacity shortfall disrupting several ongoing AI projects. While these restrictions have prompted Meta to encourage employees to use AI tokens more efficiently, reducing token usage can help lower the computational requirements.
The reported access cap comes at a time when technology companies are facing infrastructure challenges as they race to expand their AI capabilities. Despite significant investments in new data centers and AI hardware, demand for computing resources continues to outpace supply. Google itself acknowledged these constraints during its first quarter earnings announcement, where CEO Sundar Pichai noted that additional computing capacity could have led to even stronger growth for Google Cloud.
Google Cloud reported revenue of $20 billion for the quarter but also cited infrastructure constraints as a factor in a sharp increase in its order backlog. The company has been grappling with limited access to specialized chips and cloud infrastructure, which have constrained expansion efforts across the board. Other Google Cloud customers are reportedly facing similar, albeit less severe, capacity issues.
The developments highlight the ongoing competition between tech giants for AI resources. While Meta’s Gemini faces restricted access, other companies are also experiencing shortages that can impact their own AI projects and timelines. As the demand for AI continues to surge, these infrastructure challenges will likely remain a central issue in the technology industry.
Thinking about building an AI product?
Get in Touch