PALO ALTO — The scramble for computing power has become so intense that even a company like Meta, with its $200 billion market cap and sprawling data center ambitions, finds itself rationing digital tokens. Google told the social media giant around March that it could not supply the full capacity of its Gemini AI models that Meta had sought to purchase. The Financial Times broke the story.
The result: delays and disruptions to internal AI projects at Meta. This is not a story about two companies feuding.
It is a story about a wall. The wall is made of silicon, electricity, and physical space. Everyone is running into it.
Revenue at Google Cloud hit $20 billion in the first quarter ended March. CEO Sundar Pichai has said that growth would have been higher if not for computing power constraints.
Those constraints helped double the cloud unit’s backlog quarter over quarter. Google is selling what it cannot build fast enough. Meta wanted more Gemini than Google could give.
The Financial Times reports that several other Google clients have also been affected, though to a lesser extent. Meta’s demand is simply that much bigger. So the company has told its staff to become more efficient with AI tokens — the units that measure how much AI processing a task consumes.
Do more with less. That is the message inside a company that spent tens of billions last year on capital expenditures.
This is the real face of the AI boom. Not the demos. Not the breathless product launches.
The real story is the quiet, grinding battle for compute. Companies are pouring billions into chips and data centers, but the demand for AI services is growing faster than the infrastructure.
Every new model launch, every new feature, every new chatbot eats more processing power. The supply chain for advanced chips is strained. The energy grid is strained.
The timelines slip. The relationship between Google and Meta has always been complicated. They compete for advertising dollars.
They compete for talent. They compete for users.
But they also need each other. Google needs cloud customers. Meta needs computing power.
That tension is now playing out in the open, or at least in the pages of the Financial Times. Google told Meta no. That is a significant moment.
It suggests that Google is prioritizing its own AI ambitions over the needs of a major customer, or simply that it cannot keep up with anyone’s demand. For the engineers at Meta, the pressure is now personal. They are being asked to squeeze more performance out of fewer tokens.
That means rewriting code, optimizing models, and making hard choices about what gets built and what waits. The AI arms race is not just about who has the best algorithms. It is about who can get the most out of the hardware they have.
The Financial Times report makes clear that this is not a temporary hiccup. Pichai’s comments about the backlog suggest a structural problem.
The cloud unit’s backlog nearly doubled in a single quarter. That means customers are waiting longer for capacity. It means projects are being shelved.
It means the cost of AI is rising, not just in dollars but in time and opportunity. This is where the industry stands in mid-2024.
The biggest companies in the world are struggling to buy enough computing power. They are asking their teams to do more with less. And they are learning that even the most advanced technology is still bound by the limits of physics and supply chains.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Google restricts AI model usage for Meta, causing project delays”, “description”: “PALO ALTO — The scramble for computing power has become so intense that even a company like Meta, with its $200 billion market cap and…”, “datePublished”: “2026-06-28T06:05:58+00:00T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “James Roberto”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
